Communication between neurons in the brain is critical for cognition, and depends upon action potentials being conveyed along axons within white matter tracts. The speed at which these electrical signals travel along axons is known as the conduction velocity. It has been suggested that faster axonal conduction velocity promotes better cognition. For example, an increase in axonal conduction velocity is hypothesised to underpin the greater cognitive processing ability of vertebrates, in particular primates and humans (Brancucci, 2012; Miller, 1994), compared to invertebrates (Arancibia-Cárcamo et al., 2017; Bullock et al., 1984; Nave, 2010). In rats (Aston-Jones et al., 1985) and cats (Xi et al., 1999), faster axonal conduction velocity has been observed in younger compared to older animals. In a similar vein, axonal degradation that can lead to reduced conduction velocity has been identified in older compared to younger monkeys (Peters et al., 2000; Peters and Sethares, 2002).

Echoing these findings from non-humans, estimates of conduction velocity in humans from the latency of visual evoked potentials recorded over primary visual cortex (Reed and Jensen, 1992) and between the thalamus and parietal cortex (Reed and Jensen, 1993) have been positively correlated with nonverbal intelligence quotients. In addition, faster axonal conduction velocities are thought to better explain increases in intelligence compared to absolute and relative brain volumes (Dicke and Roth, 2016).

The conduction velocity of an axon is dependent upon the axon diameter, the presence and thickness of a myelin sheath, the distance between the nodes of Ranvier (periodic gaps in the myelin that facilitate action potential propagation), inter-nodal spacing, and electrical properties of the axonal and myelin membranes (Arancibia-Cárcamo et al., 2017; Drakesmith et al., 2019; Gasser and Grundfest, 1939; Hursh, 1939; Huxley and Stampfli, 1949; Rushton, 1951). A number of these features are not yet measurable in humans in vivo. However, seminal electrophysiological work has derived a relationship between axon morphology and conduction velocity using only axon diameter and myelin sheath thickness (Rushton, 1951). These two metrics are particularly key because a larger axon diameter results in less resistance to the action potential ion flow, resulting in faster conduction velocity. The presence of a myelin sheath around an axon is beneficial in two ways. First, the myelin sheath acts like an electrical insulating layer, reducing ion loss and preserving the action potential. Second, the presence of unmyelinated gaps in the myelin sheath (the nodes of Ranvier) enables a process called saltatory propagation to take place. As the majority of the axon is wrapped in myelin, the nodes of Ranvier are the only locations where action potentials can occur. This increases the strength of electrical signals because all the ions gather at these nodes instead of being dispersed along the length of the axon. Stronger action potentials are therefore sent along the myelinated portion of the axon at higher speeds, with this signal being boosted on arrival at the next node of Ranvier by another action potential, which helps to maintain a fast conduction velocity.

The conduction velocity of an axon is, therefore, not determined only by the axon diameter, but also by the relationship between the axon diameter and the thickness of the surrounding myelin sheath, a measure known as the g-ratio (Chomiak and Hu, 2009; Rushton, 1951; Schmidt and Knösche, 2019). Specifically, Rushton, 1951 derived an equation: $\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{}\mathrm{v}\mathrm{e}\mathrm{l}\mathrm{o}\mathrm{c}\mathrm{i}\mathrm{t}\mathrm{y}\mathrm{}\propto \mathrm{d}\sqrt{-\mathrm{l}\mathrm{n}\left(\mathrm{g}\right)}$ where d is the inner axon diameter, g is the g-ratio=d/D, and D is the outer fibre (axon plus myelin sheath) diameter. In other words, the g-ratio is computed as the ratio of the inner axon diameter relative to that of the axon plus the myelin sheath that encases it (Figure 1).

Figure 1

Download asset Open asset

Until recently, g-ratio measurements were restricted to invasive studies in non-human animals. However, by combining diffusion magnetic resonance imaging (MRI) with quantitative structural MRI scans optimised to assess myelination (e.g. magnetisation transfer saturation; Weiskopf et al., 2013), it is now possible to estimate the g-ratio in vivo in humans across the whole brain (Drakesmith et al., 2019; Mohammadi et al., 2015; Mohammadi and Callaghan, 2021; Stikov et al., 2015). This is achieved by measuring an aggregate g-ratio, which is an area-weighted ensemble average across a voxel of an underlying distribution of microscopic g-ratios of axons (Stikov et al., 2015; West et al., 2016). Whole brain MR g-ratio maps enable the investigation of the MR g-ratio of white matter fibre pathways at the group level. These MR g-ratio estimates have been optimised (Ellerbrock and Mohammadi, 2018; Jung et al., 2018; West et al., 2018) and used to investigate white matter development (Cercignani et al., 2017), changes in the g-ratio during aging (Berman et al., 2018), and as a potential neuroimaging marker in patients with multiple sclerosis (Yu et al., 2019). Of note, the MR g-ratio has been found to associate well with estimates of axonal conduction velocity (Berman et al., 2019; Drakesmith et al., 2019). Consequently, the MR g-ratio provides a non-invasive MRI method that can associate in vivo structural neuroimaging of humans with axonal conduction velocity.

Given that the MR g-ratio encompasses information about the inner axon diameter and myelin thickness, the identification of statistically significant relationships with the MR g-ratio can also guide as to which of these two features might be influencing variations in conduction velocity (Caeyenberghs et al., 2016; Kaller et al., 2017; Lakhani et al., 2016; Waxman, 1980; Xin and Chan, 2020). This is particularly useful in relation to the inner axon diameter. While measures of myelination can be obtained using conventional MRI scanners (e.g. via magnetisation transfer saturation maps), estimates of inner axon diameter cannot, and instead require ultra-strong gradient systems – 300 mT/m compared to the 10s of mT/m of conventional MRI scanners (Jones et al., 2018; Veraart et al., 2020). However, only four ultra-strong gradient Connectom MRI scanners exist in the world. Consequently, measuring the MR g-ratio with conventional MRI scanners can be useful in guiding inferences about inner axon diameter in the context of conduction velocity.

Adjudicating between the possible influence of inner axon diameter or myelination relies on knowledge, or an assumption, about whether the associated change in conduction velocity is faster or slower. This is because significant associations with the MR g-ratio only indicate the existence of a relationship with conduction velocity but not the direction. As noted previously, faster conduction velocity is often held to promote better cognition (e.g. Brancucci, 2012; Dicke and Roth, 2016; Miller, 1994; Reed and Jensen, 1992). Therefore, in Figure 2A, inner axon diameter, myelin thickness and (MR) g-ratio are plotted together to illustrate how different changes in the MR g-ratio are related to the underlying microstructural properties, given a faster conduction velocity. Myelin thickness is represented by the gradient in background colour and contours, with thinnest myelin at the bottom right, and thickest on the top left. The direction of the arrows describes the change in g-ratio for the microstructural variations presented. There are three main scenarios. (1) A decrease in MR g-ratio values. This would suggest that faster conduction velocity is due to greater thickness of the myelin sheath, with the inner axon diameter remaining constant (Figure 2A blue arrow, and Figure 2B). (2) A decrease in MR g-ratio values, but to a lesser extent than that observed in the first scenario. This would suggest that faster conduction velocity is due primarily to greater myelin sheath thickness, but one that is also accompanied by a larger inner axon diameter (Figure 2A red arrow, and Figure 2C). (3) An increase in MR g-ratio values. This would suggest that faster conduction velocity is predominantly due to a larger inner axon diameter, with only small differences, if any, in myelin thickness being present (Figure 2A orange arrow, and Figure 2D).

Figure 2

Download asset Open asset

We note, for completeness, that a fourth scenario also exists where constant MR g-ratio values could be associated with faster conduction velocity. This would occur when both the inner axon diameter and myelin thickness change proportionally to each other (Figure 2A black arrow, and Figure 2E). However, this scenario could also mean that there is no variation in conduction velocity. One way to increase interpretability in this situation is by examining scans optimised to assess myelination (such as magnetisation transfer saturation values). Observing no relationship with the MR g-ratio, but a relationship with magnetisation transfer saturation values, would suggest a proportional change in the underlying myelin and, consequently, variation in conduction velocity. By contrast, no relationship with either the MR g-ratio or magnetisation transfer saturation values would suggest that there was no change in the underlying microstructure and therefore no variation in conduction velocity.

The MR g-ratio has, therefore, the potential to provide a number of novel insights into human cognition. One area where the MR g-ratio may be particularly helpful is in probing individual differences. Our particular interest is in the ability to recall past experiences from real life, known as autobiographical memories. The detailed recall of autobiographical memories is a critical cognitive function that serves to sustain our sense of self, enable independent living, and prolong survival (Tulving, 2002). While some healthy individuals can recollect decades-old autobiographical memories with great richness and clarity, others struggle to recall what they did last weekend (LePort et al., 2012; Palombo et al., 2015). In the context of the healthy population, we currently lack a clear biological explanation for the basis of these individual differences (Palombo et al., 2018).

There is no doubt that the hippocampus is central to the processing of autobiographical memories, and hippocampal damage is linked with autobiographical memory impairments (McCormick et al., 2018; Scoville and Milner, 1957; Winocur and Moscovitch, 2011). However, no consistent relationship between autobiographical memory recall ability and hippocampal grey matter volume or microstructure has been identified in healthy individuals (Clark et al., 2020; Clark et al., 2021a; LePort et al., 2012; Maguire et al., 2003; Van Petten, 2004). The hippocampus does not act alone, and functional neuroimaging studies have revealed that a distributed set of brain areas supports autobiographical memory recall along with the hippocampus, including the parahippocampal, retrosplenial, parietal and medial prefrontal cortices (Andrews-Hanna et al., 2014; Maguire, 2001; Spreng et al., 2009; Svoboda et al., 2006). Changes in MR g-ratio and, by inference, conduction velocity, might affect communication between these brain regions and so influence individual differences in autobiographical memory recall within the healthy population.

Three white matter pathways in particular enable communication with the hippocampal region – the fornix, the uncinate fasciculus and the parahippocampal cingulum bundle. The fornix (Figure 3A) is a major pathway in and out of the hippocampus and connects it to the orbital and medial prefrontal cortices, the basal forebrain, the anterior thalamus, the hypothalamus and the mammillary bodies (Aggleton et al., 2015; Croxson et al., 2005). The uncinate fasciculus (Figure 3B) originates in the uncus, entorhinal and parahippocampal cortices and passes over the lateral nucleus of the amygdala, arcs around the Sylvian fissure, terminating in various locations throughout the prefrontal cortex (Croxson et al., 2005; Von Der Heide et al., 2013). The parahippocampal cingulum bundle (Figure 3C) links the hippocampus with the entorhinal, parahippocampal, retrosplenial and parietal cortices, as well as providing another route between the hippocampus and anterior thalamus (Bubb et al., 2018; Jones et al., 2013b). It also links to the prefrontal cortex via its connections with other parts of the cingulum bundle.

Figure 3

Download asset Open asset

In the current study, we calculated the MR g-ratio within these three pathways to ascertain whether this was related to autobiographical memory recall ability. A significant relationship with the MR g-ratio would suggest that variation in the conduction velocity of a pathway is associated with autobiographical memory recall ability. The relationship of MR g-ratio to the underlying microstructure as outlined in Figure 2, would further guide us as to whether any significant effects were more likely to be associated with the extent of myelination or the size of the inner axon diameter of the fibres in these three white matter tracts. These analyses were augmented by examining whether magnetisation transfer saturation values of the pathways (assessing myelination) were associated with autobiographical memory recall ability.

In addition, we also used the neurite orientation dispersion and density imaging (NODDI; Zhang et al., 2012) biophysical model to derive two complementary biological measures that could provide further insights into the arrangement of neurites in a voxel. A neurite is any projection from a neuron’s cell body, such as an axon or a dendrite. The neurite orientation dispersion index is an estimate of the organisation of the neurites in a voxel, where a small orientation dispersion index value indicates a low dispersion of neurites, in other words, that the neurites are coherently organised. The second property, neurite density, is a measure of the density of the neurites in a voxel.

For completeness, the commonly reported physical parameters from standard diffusion tensor imaging (DTI; e.g. fractional anisotropy and mean diffusivity; Basser, 1995) that are often derived from diffusion data were also computed (Oeschger et al., 2021). However, these metrics lack biological specificity (Jensen and Helpern, 2010; Jones et al., 2013a) and, consequently, could not speak to our research questions. For example, fractional anisotropy is a very general measure. Variations in fractional anisotropy values can occur for numerous reasons, including, but not limited to, changes in the extent of myelination, axon coherence, axon density, and the level of astrocytes. Consequently, standard DTI parameter results are only briefly summarised in the main text, with full details provided in Appendix 1 and Supplementary file 1, Supplementary file 2.

To ensure an appropriate sample size and a wide range of autobiographical memory recall ability, we examined a large group of healthy young adults from the general population (n=217; 109 females, 108 males; mean age of 29.0 years, SD = 5.60; age was restricted to between 20 and 41 years to limit the possible effects of aging).

All participants underwent the widely-used Autobiographical Interview (Levine et al., 2002). This task, which is the gold standard in the field, provides a detailed metric characterising a person’s ability to recall real-life past experiences, as well as providing a useful control measure. We also examined performance on another eight standard laboratory-based memory tests. Their inclusion allowed us to ascertain whether any relationships with conduction velocity were specific to recollecting detailed autobiographical memories from real life, or were also applicable to the recall of more constrained laboratory-based stimuli.

Diffusion and magnetisation transfer saturation MRI scans were obtained for each person to enable calculation of the MR g-ratio and the other measures. Our analyses were performed using weighted means from each of the three white matter tracts of interest rather than voxel-wise across the whole brain, reducing the potential for false positives (Marek et al., 2022).

Focusing on the fornix, uncinate fasciculus and parahippocampal cingulum bundle, we predicted that variations in the MR g-ratio from some or all of these tracts would be associated with autobiographical memory recall ability. Such a finding would, for the first time, suggest a link between variations in white matter tract conduction velocity and individual differences in autobiographical memory recall.

Under the assumption that a significant relationship between the MR g-ratio and autobiographical memory recall represents an association with faster conduction velocity (e.g. Brancucci, 2012; Dicke and Roth, 2016; Miller, 1994; Reed and Jensen, 1992), and aided by the analysis of the magnetisation transfer saturation values, we further sought to evaluate the scenarios presented in Figure 2. Specifically, we asked whether autobiographical memory recall ability was more likely to be associated with inner axon diameter or myelin thickness in the context of g-ratio and conduction velocity. A negative relationship between the MR g-ratio and autobiographical memory recall ability (Figure 2A blue and red arrows, and Figure 2B and C), along with a positive relationship between magnetisation transfer saturation and autobiographical memory recall ability, would suggest that myelin thickness was more relevant. By contrast, a positive relationship between the MR g-ratio and autobiographical memory recall ability (Figure 2A orange arrow, and Figure 2D) would highlight the potential relevance of inner axon diameters. Observing no associations between the MR g-ratio or magnetisation transfer saturation values and autobiographical memory recall, would suggest no relationships with the underlying microstructure or conduction velocity, and would speak against the possibility of a proportional change in microstructure (Figure 2A black arrow, and Figure 2E).

Finally, given that a previous functional MRI meta-analysis identified different neural substrates associated with autobiographical memory recall and the recall of laboratory-based stimuli (McDermott et al., 2009; see also Maguire, 2001; Maguire, 2012; Miller et al., 2022; Mobbs et al., 2021; Nastase et al., 2020; Spiers and Maguire, 2007), we expected that associations with the MR g-ratio might be specific to autobiographical memory recall. We reasoned that detailed, multimodal, autobiographical memories may rely on inter-regional connectivity to a greater degree than simpler, more constrained laboratory-based memory tests.

The conduction velocity of action potentials along axons is crucial for neural communication. Until recently it was not possible to examine metrics associated with axonal conduction velocity, such as the g-ratio, in vivo in the human brain, with this being largely the preserve of studies involving non-human animals. However, by combining diffusion MRI with quantitative structural MRI scans optimised to assess myelination (Callaghan et al., 2014), it is now possible to estimate the MR g-ratio in vivo in humans (Drakesmith et al., 2019; Mohammadi et al., 2015; Mohammadi and Callaghan, 2021; Stikov et al., 2015). Here, in the first application to human cognition, we found that variations in the MR g-ratio specifically in the parahippocampal cingulum bundle were associated with individual differences in autobiographical memory recall ability in a large sample of healthy adults. Under various modelling assumptions related to MR g-ratio and conduction velocity (detailed below), this was associated with larger inner axon diameters rather than myelin content. Moreover, we identified a possible link between autobiographical memory recall ability and more coherently organised neurites. Our findings were also specific to autobiographical memory retrieval and were not evident for laboratory-based memory tests. These results offer a new perspective on individual differences in autobiographical memory recall ability. This is especially welcome given the current lack of a clear biological explanation for such variations in the healthy population (Clark et al., 2020; Clark et al., 2021a; LePort et al., 2012; Maguire et al., 2003; Palombo et al., 2018; Van Petten, 2004).

A key property of the MR g-ratio is that it provides a non-invasive methodology with which to relate in vivo structural neuroimaging of humans with axonal conduction velocity (Berman et al., 2019; Drakesmith et al., 2019). Moreover, the identification of a statistically significant relationship with the MR g-ratio can help to guide as to which of inner axon diameter or myelin thickness is more likely to be associated with variations in conduction velocity (Caeyenberghs et al., 2016; Kaller et al., 2017; Lakhani et al., 2016; Waxman, 1980; Xin and Chan, 2020).

Our inferences rested upon the assumption that a significant relationship with the MR g-ratio reflected faster conduction velocity, given previous work suggesting that faster conduction velocity might be associated with better cognition (e.g. Brancucci, 2012; Dicke and Roth, 2016; Miller, 1994; Reed and Jensen, 1992). As shown in Figure 2, there are three main scenarios describing how changes in the MR g-ratio are related to the underlying microstructural properties, given a faster conduction velocity. The positive relationship we observed between the parahippocampal cingulum bundle MR g-ratio and autobiographical memory recall ability suggests that this effect was associated predominantly with larger inner axon diameter (Figure 2A orange arrow, and Figure 2D). By contrast, had a negative correlation between the MR g-ratio and autobiographical memory recall ability been identified, we could instead have inferred that myelination was the relevant microstructural feature. The lack of relationship between parahippocampal cingulum bundle magnetisation transfer saturation values (optimised to assess myelination) and autobiographical memory recall ability provided further corroboration that myelin may not have been influential in this context. Greater myelination is often held to be a prominent influence on behavioural and cognitive performance (Caeyenberghs et al., 2016; Fields and Bukalo, 2020; Kaller et al., 2017; Lakhani et al., 2016; Waxman, 1980; Xin and Chan, 2020). By contrast, our results highlight the potentially important role that the inner axon diameter could be playing in autobiographical memory recall.

We also found that lower neurite dispersion, suggesting more coherent neurite organisation, was related to better autobiographical memory recall ability. Our measure of neurite dispersion was obtained using the NODDI biophysical model (Zhang et al., 2012), which aims to isolate the organisation of the neurites in a voxel from the density of the neurites in a voxel. While neurite dispersion was significantly related to autobiographical memory recall, no relationship was observed with neurite density.

A larger inner axon diameter reduces resistance to action potential signals, enabling greater conduction velocities (Gasser and Grundfest, 1939; Hursh, 1939), and coherently organised fibres may decrease the distance signals need to travel, further reducing communication times (Salami et al., 2003; Steriade, 1995). This combination of features might optimise a fibre bundle for faster communication, and this may benefit autobiographical memory recall.

These results were specific to one white matter tract, the parahippocampal cingulum bundle (Bubb et al., 2018). MR g-ratio measures from no other tract, including the dorsal part of the cingulum bundle, showed any association with autobiographical memory recall ability. The parahippocampal cingulum bundle directly connects the hippocampus, parahippocampal, retrosplenial and parietal cortices, the anterior thalamus and, through other subdivisions of the cingulum bundle, the medial prefrontal cortex. These regions are typically engaged during fMRI studies of autobiographical memory recall (Andrews-Hanna et al., 2014; Maguire, 2001; Spreng et al., 2009; Svoboda et al., 2006), and damage to them is often associated with autobiographical memory impairments (Berryhill et al., 2007; McCormick et al., 2018; Scoville and Milner, 1957; Vann et al., 2009). The retrosplenial and parahippocampal cortices are thought to provide visuospatial elements of autobiographical memories (Dalton and Maguire, 2017; Epstein and Higgins, 2007; Mullally and Maguire, 2011; Vann et al., 2009), while the medial prefrontal cortex may initiate autobiographical retrieval and support schema-guided recall (Gilboa and Marlatte, 2017; McCormick et al., 2018; McCormick et al., 2020). The parahippocampal cingulum bundle is, therefore, uniquely positioned as a transmission highway enabling this information to reach the hippocampus, where memories can be reconstructed (Bartlett, 1932; Hassabis and Maguire, 2007; Schacter et al., 2012). Larger inner axon diameters and coherently organised neurites could facilitate rapid information flow along the parahippocampal cingulum bundle leading to more memory elements being available simultaneously, which in turn may result in increased detail and better integration of a memory representation.

There have only been a small number of previous studies investigating white matter tracts and individual differences in autobiographical memory recall ability in healthy people. These focused on standard DTI parameters, such as fractional anisotropy and mean diffusivity, and not the MR g-ratio. Two such studies were suggestive of a relationship between parahippocampal cingulum bundle fractional anisotropy and autobiographical memory recall (Irish et al., 2014; Memel et al., 2020). Our analyses of the standard DTI parameters, therefore, replicated these previous results. Direct comparison between these studies and our findings is difficult, however, given the relatively small sample sizes, older age of participants in some of the extant studies, and the testing of mixed groups of healthy older people and patients with dementia. In addition, fractional anisotropy and other physical parameters lack biological specificity (Jensen and Helpern, 2010; Jones et al., 2013a) and, consequently, they cannot speak to questions concerning axonal conduction velocity, which is our main interest here. In a similar vein, we are not aware of any reports of bilateral lesions that selectively compromise or sever the parahippocampal cingulum bundle in humans. Given our findings, we would predict that such lesions would adversely affect the ability to recall autobiographical memories.

Despite its prime location and connectivity at the heart of the brain’s autobiographical memory system, the dearth of studies in humans and the rarity of selective bilateral lesions to the parahippocampal cingulum bundle have perhaps obscured its importance when compared to other, more celebrated, memory-related white matter tracts. We also examined two such tracts, the fornix and the uncinate fasciculus, but in both cases no significant relationships between any of our neuroimaging metrics and autobiographical memory recall ability were evident in our large cohort of young healthy adults. As outlined in Figure 2, there was one possible scenario where a null relationship with the MR g-ratio could reflect an association with conduction velocity – if both the inner axon diameter and myelin thickness increased proportionally to each other (Figure 2A black arrow, and Figure 2E). However, we found no associations between autobiographical memory recall and both the MR g-ratio and magnetisation transfer saturation values of the fornix and uncinate fasciculus, which speaks against this explanation. In terms of standard DTI parameters, unlike a previous study (Hodgetts et al., 2017), we did not find a relationship between fornix fractional anisotropy and autobiographical memory recall ability, which may be due to the larger sample that we examined.

Both the fornix and parahippocampal cingulum bundle are vulnerable to partial volume effects (Concha et al., 2005). However, to mitigate this issue we took steps to ensure the data were extracted only from white matter voxels (see Materials and methods). Diffusion data are also susceptible to distortions which can particularly affect the uncinate fasciculus. We addressed this challenge by using a new technique that improves distortion correction in this region (Clark et al., 2021b). The absence of fornix findings in our study could echo those relating to hippocampal volume (Clark et al., 2020; Clark et al., 2021a), whereby the structure is widely acknowledged to be involved in autobiographical memory recall, and damage impedes retrieval (Aggleton et al., 2000; D’Esposito et al., 1995; Gaffan and Gaffan, 1991; Tsivilis et al., 2008), but microstructural variations have limited impact within the healthy adult population. Regarding the uncinate fasciculus, unilateral lesions do not seem to significantly impair performance on laboratory-based memory tests (Papagno et al., 2011) or the recall of premorbid autobiographical memories (Levine et al., 2009). Bilateral uncinate fasciculus lesions are very rare in humans, but might result in greater memory impairment. Alternatively, hippocampal-prefrontal connections may be better served by other pathways, for example, via the fornix or parts of the cingulum bundle.

Our findings were not only specific to the parahippocampal cingulum bundle, but also to the internal details of autobiographical memories, which reflect the episodicity of past experiences. By contrast, our control measure of external details, also from the same task but which concerns non-episodic information in the autobiographical memories, did not correlate with any white matter microstructure metrics. Moreover, no relationships were identified with any of the eight laboratory-based memory tasks examined, highlighting the specific nature of the relationship between parahippocampal cingulum bundle MR g-ratio and, by inference, conduction velocity, and autobiographical memory recall ability. It may be that vivid, detailed, multimodal, autobiographical memories rely on inter-regional connectivity, particularly that supported by the parahippocampal cingulum bundle, to a greater degree than simpler, more constrained laboratory-based memory tests. This result aligns with previous work involving an fMRI meta-analysis which showed that recall of autobiographical memories and laboratory-based memory stimuli were associated with substantially different neural substrates (McDermott et al., 2009). More generally, our findings add to the increasingly-recognised importance of studying real-world cognition in order to fully characterise brain-behaviour relationships (Maguire, 2001; Maguire, 2012; Miller et al., 2022; Mobbs et al., 2021; Nastase et al., 2020; Spiers and Maguire, 2007).

Age, gender, scanner and the number of voxels within an ROI were included as covariates in all analyses, limiting the potential confounding effects of these variables. Our analyses were also performed using weighted means from each of the three white matter tracts of interest rather than voxel-wise across the whole brain, reducing the potential for false positives (Marek et al., 2022). We also tested a sample of over 200 participants, which has been suggested as sufficient for correlational neuroimaging research such as that performed here (Cecchetti and Handjaras, 2022; DeYoung et al., 2022). Relationships with autobiographical memory recall ability were also specific to the MR g-ratio, and were not evident for any of its components (magnetisation transfer saturation, neurite density or isotropic volume fraction), suggesting that it is the combination of these measures, in the form of the MR g-ratio, that is meaningful rather than any single measure alone.

Nevertheless, as with all neuroimaging techniques, methodological limitations need to be considered when measuring the MR g-ratio (see Campbell et al., 2018; Mohammadi and Callaghan, 2021 for in depth methodological reviews). First, the MR g-ratio is an area-weighted average of all microscopic axons in an MRI-voxel that is slightly weighted towards larger axons. Second, MR proxies are required to estimate myelin and axonal volumes for the calculation of the MR g-ratio. Multiple methodologies are available to estimate myelin and axonal volumes, with no consensus yet reached as to the best combination (Mohammadi and Callaghan, 2021). Third, a calibration step is required to more closely align the MR proxies with the estimated volume fractions. Here, we used the standard single-point calibration method to estimate the slope of the myelin-based proxy, assuming that the offset can be neglected. Future work may be able to improve these calibrations. But, of note, since a non-negligible offset in the myelin-based proxy can increase the error in the MR g-ratio, any calibration improvements would likely serve to increase the observed correlation. Fourth, although the model of the MR g-ratio allows for fibre dispersion (Stikov et al., 2015), this is not accounted for in mapping to conduction velocity. Going forward, more advanced models may help to further elaborate on these relationships. Fifth, our myelin measure can be influenced by any factor, cognitive or demographic, that leads to a difference in absolute myelination. In principle, unexplained variance could obscure a true underlying relationship with autobiographical memory. To mitigate this, our analyses controlled for age, gender and any potential scanner-related differences. The MR g-ratio, however, is less affected by this limitation because it is a relative measure and does not depend on absolute myelin content but on the balance between absolute myelin and axonal volumes.

Finally, our inference suggesting that the positive correlation between the MR g-ratio and autobiographical memory recall ability is potentially related to inner axon diameter is specifically in the context of g-ratio and conduction velocity, and is based upon Rushton’s (1951) model rather than direct measurements of axon diameter. The use of Connectom MRI scanners (Jones et al., 2018) and more advanced biophysical models may be able to expand upon this relationship in the future.

While we examined the parahippocampal cingulum bundle as a unitary pathway, it comprises both long and short association fibres with differing connectivity (Bubb et al., 2018). Some fibres will form long range connections between, for example, the hippocampus and retrosplenial cortex, whereas others will make shorter range connections between neighbouring regions. Connectom MRI is starting to make examination of short range ‘u-fibres’ possible in vivo in humans (Movahedian Attar et al., 2020; Shastin et al., 2022). While there are currently only four of these scanners in the world, future studies using Connectom MRI could seek to identify specific connections within the parahippocampal cingulum bundle that relate to individual differences in autobiographical memory recall.

In conclusion, white matter microstructure measures related to conduction velocity are now possible to derive in vivo in humans. This has the potential to provide novel insights into how the brain processes and integrates information (Berman et al., 2019; Drakesmith et al., 2019), deepening our understanding of the information flow that underpins critical cognitive functions such as autobiographical memory recall.

The data for every participant are provided in Figure 5—source data 1, Figure 6—source data 1, Supplementary file 1 and Supplementary file 2.

1. 1. Aggleton JP
   2. McMackin D
   3. Carpenter K
   4. Hornak J
   5. Kapur N
   6. Halpin S
   7. Wiles CM
   8. Kamel H
   9. Brennan P
   10. Carton S
   11. Gaffan D

   (2000) Differential cognitive effects of colloid cysts in the third ventricle that spare or compromise the fornix

   Brain 123 (Pt 4):800–815.

   https://doi.org/10.1093/brain/123.4.800

   • PubMed
   • Google Scholar
2. 1. Aggleton JP
   2. Wright NF
   3. Rosene DL
   4. Saunders RC

   (2015) Complementary patterns of direct amygdala and hippocampal projections to the macaque prefrontal cortex

   Cerebral Cortex 25:4351–4373.

   https://doi.org/10.1093/cercor/bhv019

   • PubMed
   • Google Scholar
3. 1. André ED
   2. Grinberg F
   3. Farrher E
   4. Maximov II
   5. Shah NJ
   6. Meyer C
   7. Jaspar M
   8. Muto V
   9. Phillips C
   10. Balteau E

   (2014) Influence of noise correction on intra- and inter-subject variability of quantitative metrics in diffusion kurtosis imaging

   PLOS ONE 9:e94531.

   https://doi.org/10.1371/journal.pone.0094531

   • PubMed
   • Google Scholar
4. 1. Andrews-Hanna JR
   2. Saxe R
   3. Yarkoni T

   (2014) Contributions of episodic retrieval and mentalizing to autobiographical thought: evidence from functional neuroimaging, resting-state connectivity, and fMRI meta-analyses

   NeuroImage 91:324–335.

   https://doi.org/10.1016/j.neuroimage.2014.01.032

   • PubMed
   • Google Scholar
5. 1. Arancibia-Cárcamo IL
   2. Ford MC
   3. Cossell L
   4. Ishida K
   5. Tohyama K
   6. Attwell D

   (2017) Node of Ranvier length as a potential regulator of myelinated axon conduction speed

   eLife 6:e23329.

   https://doi.org/10.7554/eLife.23329

   • PubMed
   • Google Scholar
6. 1. Ashburner J
   2. Friston KJ

   (2005) Unified segmentation

   NeuroImage 26:839–851.

   https://doi.org/10.1016/j.neuroimage.2005.02.018

   • Google Scholar
7. 1. Ashburner J

   (2007) A fast diffeomorphic image registration algorithm

   NeuroImage 38:95–113.

   https://doi.org/10.1016/j.neuroimage.2007.07.007

   • PubMed
   • Google Scholar
8. 1. Aston-Jones G
   2. Rogers J
   3. Shaver RD
   4. Dinan TG
   5. Moss DE

   (1985) Age-impaired impulse flow from nucleus basalis to cortex

   Nature 318:462–464.

   https://doi.org/10.1038/318462a0

   • PubMed
   • Google Scholar
9. Book

   1. Bartlett FC

   (1932)

   Remembering: A Study in Experimental and Social Psychology

   Cambridge University Press.

   • Google Scholar
10. 1. Basser PJ

    (1995) Inferring microstructural features and the physiological state of tissues from diffusion-weighted images

    NMR in Biomedicine 8:333–344.

    https://doi.org/10.1002/nbm.1940080707

    • PubMed
    • Google Scholar
11. 1. Becker SMA
    2. Tabelow K
    3. Voss HU
    4. Anwander A
    5. Heidemann RM
    6. Polzehl J

    (2012) Position-orientation adaptive smoothing of diffusion weighted magnetic resonance data (POAS)

    Medical Image Analysis 16:1142–1155.

    https://doi.org/10.1016/j.media.2012.05.007

    • PubMed
    • Google Scholar
12. 1. Berman S
    2. West KL
    3. Does MD
    4. Yeatman JD
    5. Mezer AA

    (2018) Evaluating g-ratio weighted changes in the corpus callosum as a function of age and sex

    NeuroImage 182:304–313.

    https://doi.org/10.1016/j.neuroimage.2017.06.076

    • Google Scholar
13. 1. Berman S
    2. Filo S
    3. Mezer AA

    (2019) Modeling conduction delays in the corpus callosum using MRI-measured g-ratio

    NeuroImage 195:128–139.

    https://doi.org/10.1016/j.neuroimage.2019.03.025

    • Google Scholar
14. 1. Berryhill ME
    2. Phuong L
    3. Picasso L
    4. Cabeza R
    5. Olson IR

    (2007) Parietal lobe and episodic memory: bilateral damage causes impaired free recall of autobiographical memory

    Journal of Neuroscience 27:14415.

    https://doi.org/10.1523/JNEUROSCI.4163-07.2007

    • Google Scholar
15. 1. Brancucci A

    (2012) Neural correlates of cognitive ability

    Journal of Neuroscience Research 90:1299–1309.

    https://doi.org/10.1002/jnr.23045

    • Google Scholar
16. 1. Bubb EJ
    2. Metzler-Baddeley C
    3. Aggleton JP

    (2018) The cingulum bundle: anatomy, function, and dysfunction

    Neuroscience and Biobehavioral Reviews 92:104–127.

    https://doi.org/10.1016/j.neubiorev.2018.05.008

    • PubMed
    • Google Scholar
17. 1. Bullock TH
    2. Moore JK
    3. Fields RD

    (1984) Evolution of myelin sheaths: both lamprey and hagfish lack myelin

    Neuroscience Letters 48:145–148.

    https://doi.org/10.1016/0304-3940(84)90010-7

    • Google Scholar
18. 1. Caeyenberghs K
    2. Metzler-Baddeley C
    3. Foley S
    4. Jones DK

    (2016) Dynamics of the human structural connectome underlying working memory training

    Journal of Neuroscience 36:4056.

    https://doi.org/10.1523/JNEUROSCI.1973-15.2016

    • Google Scholar
19. 1. Callaghan MF
    2. Freund P
    3. Draganski B
    4. Anderson E
    5. Cappelletti M
    6. Chowdhury R
    7. Diedrichsen J
    8. FitzGerald THB
    9. Smittenaar P
    10. Helms G
    11. Lutti A
    12. Weiskopf N

    (2014) Widespread age-related differences in the human brain microstructure revealed by quantitative magnetic resonance imaging

    Neurobiology of Aging 35:1862–1872.

    https://doi.org/10.1016/j.neurobiolaging.2014.02.008

    • Google Scholar
20. 1. Callaghan MF
    2. Josephs O
    3. Herbst M
    4. Zaitsev M
    5. Todd N
    6. Weiskopf N

    (2015) An evaluation of prospective motion correction (PMC) for high resolution quantitative MRI

    Frontiers in Neuroscience 9:97.

    https://doi.org/10.3389/fnins.2015.00097

    • Google Scholar
21. 1. Callaghan MF
    2. Lutti A
    3. Ashburner J
    4. Balteau E
    5. Corbin N
    6. Draganski B
    7. Helms G
    8. Kherif F
    9. Leutritz T
    10. Mohammadi S
    11. Phillips C
    12. Reimer E
    13. Ruthotto L
    14. Seif M
    15. Tabelow K
    16. Ziegler G
    17. Weiskopf N

    (2019) Example dataset for the hmri toolbox

    Data in Brief 25:104132.

    https://doi.org/10.1016/j.dib.2019.104132

    • Google Scholar
22. 1. Campbell JSW
    2. Leppert IR
    3. Narayanan S
    4. Boudreau M
    5. Duval T
    6. Cohen-Adad J
    7. Pike GB
    8. Stikov N

    (2018) Promise and pitfalls of g-ratio estimation with MRI

    NeuroImage 182:80–96.

    https://doi.org/10.1016/j.neuroimage.2017.08.038

    • Google Scholar
23. Preprint

    1. Cecchetti L
    2. Handjaras G

    (2022) Reproducible brain-wide association studies do not necessarily require thousands of individuals

    PsyArXiv.

    https://doi.org/10.31234/osf.io/c8xwe

    • Google Scholar
24. 1. Cercignani M
    2. Giulietti G
    3. Dowell NG
    4. Gabel M
    5. Broad R
    6. Leigh PN
    7. Harrison NA
    8. Bozzali M

    (2017) Characterizing axonal myelination within the healthy population: a tract-by-tract mapping of effects of age and gender on the fiber g-ratio

    Neurobiology of Aging 49:109–118.

    https://doi.org/10.1016/j.neurobiolaging.2016.09.016

    • Google Scholar
25. 1. Chomiak T
    2. Hu B

    (2009) What is the optimal value of the g-ratio for myelinated fibers in the rat CNS? A theoretical approach

    PLOS ONE 4:e7754.

    https://doi.org/10.1371/journal.pone.0007754

    • Google Scholar
26. 1. Clark IA
    2. Monk AM
    3. Hotchin V
    4. Pizzamiglio G
    5. Liefgreen A
    6. Callaghan MF
    7. Maguire EA

    (2020) Does hippocampal volume explain performance differences on hippocampal-dependent tasks

    NeuroImage 221:117211.

    https://doi.org/10.1016/j.neuroimage.2020.117211

    • Google Scholar
27. 1. Clark IA
    2. Callaghan MF
    3. Weiskopf N
    4. Maguire EA

    (2021a) The relationship between hippocampal-dependent task performance and hippocampal grey matter myelination and iron content

    Brain and Neuroscience Advances 5:1–8.

    https://doi.org/10.1177/23982128211011923

    • Google Scholar
28. 1. Clark IA
    2. Callaghan MF
    3. Weiskopf N
    4. Maguire EA
    5. Mohammadi S

    (2021b) Reducing susceptibility distortion related image blurring in diffusion MRI EPI data

    Frontiers in Neuroscience 15:706473.

    https://doi.org/10.3389/fnins.2021.706473

    • PubMed
    • Google Scholar
29. 1. Cohen J

    (1992) A power primer psychological

    Bulletin 112:155–159.

    https://doi.org/10.1037/0033-2909.112.1.155

    • Google Scholar
30. 1. Concha L
    2. Gross DW
    3. Beaulieu C

    (2005)

    Diffusion tensor tractography of the limbic system

    American Journal of Neuroradiology 26:2267–2274.

    • PubMed
    • Google Scholar
31. 1. Corbin N
    2. Callaghan MF

    (2021) Imperfect spoiling in variable FLIP angle T1 mapping at 7T: quantifying and minimising impact

    Magnetic Resonance in Medicine 86:693–708.

    https://doi.org/10.1002/mrm.28720

    • Google Scholar
32. 1. Croxson PL
    2. Johansen-Berg H
    3. Behrens TEJ
    4. Robson MD
    5. Pinsk MA
    6. Gross CG
    7. Richter W
    8. Richter MC
    9. Kastner S
    10. Rushworth MFS

    (2005) Quantitative investigation of connections of the prefrontal cortex in the human and macaque using probabilistic diffusion tractography

    The Journal of Neuroscience 25:8854–8866.

    https://doi.org/10.1523/JNEUROSCI.1311-05.2005

    • PubMed
    • Google Scholar
33. 1. Dalton MA
    2. Maguire EA

    (2017) The pre/parasubiculum: a hippocampal hub for scene-based cognition?

    Current Opinion in Behavioral Sciences 17:34–40.

    https://doi.org/10.1016/j.cobeha.2017.06.001

    • PubMed
    • Google Scholar
34. 1. D’Esposito M
    2. Verfaellie M
    3. Alexander MP
    4. Katz DI

    (1995) Amnesia following traumatic bilateral fornix transection

    Neurology 45:1546.

    https://doi.org/10.1212/WNL.45.8.1546

    • Google Scholar
35. Preprint

    1. DeYoung CG
    2. Sassenberg TA
    3. Abend R
    4. Allen T
    5. Beaty R
    6. Bellgrove M
    7. Blain SD
    8. Bzdok D
    9. Chavez RS
    10. Engel SA
    11. Ma F
    12. Fornito A
    13. Genç E
    14. Goghari V
    15. Grazioplene R
    16. Hanson JL
    17. Haxby J
    18. Hilger K
    19. Homan P
    20. Joyner K
    21. Kaczkurkin A
    22. Latzman RD
    23. Martin EA
    24. Passamonti L
    25. Pickering A
    26. Safron A
    27. Servaas M
    28. Smillie L
    29. Spreng RN
    30. Tiego J
    31. Viding E
    32. Wacker J

    (2022) Reproducible Between-Person Brain-Behavior Associations Do Not Always Require Thousands of Individuals

    PsyArXiv.

    https://doi.org/10.31234/osf.io/sfnmk

    • Google Scholar
36. 1. Dicke U
    2. Roth G

    (2016) Neuronal factors determining high intelligence

    Philosophical Transactions of the Royal Society of LondonSeries B: Biological Sciences 371:20150180.

    https://doi.org/10.1098/rstb.2015.0180

    • Google Scholar
37. 1. Diedenhofen B
    2. Musch J

    (2015) Cocor: a comprehensive solution for the statistical comparison of correlations

    PLOS ONE 10:e0121945.

    https://doi.org/10.1371/journal.pone.0121945

    • Google Scholar
38. 1. Drakesmith M
    2. Harms R
    3. Rudrapatna SU
    4. Parker GD
    5. Evans CJ
    6. Jones DK

    (2019) Estimating axon conduction velocity in vivo from microstructural

    MRI NeuroImage 203:116186.

    https://doi.org/10.1016/j.neuroimage.2019.116186

    • Google Scholar
39. 1. Ellerbrock I
    2. Mohammadi S

    (2018) Four in vivo g-ratio-weighted imaging methods: comparability and repeatability at the group level

    Frontiers in Neuroscience 39:24–41.

    https://doi.org/10.1002/hbm.23858

    • Google Scholar
40. 1. Epstein RA
    2. Higgins JS

    (2007) Differential parahippocampal and retrosplenial involvement in three types of visual scene recognition

    Cerebral Cortex 17:1680–1693.

    https://doi.org/10.1093/cercor/bhl079

    • PubMed
    • Google Scholar
41. 1. Feinberg DA
    2. Moeller S
    3. Smith SM
    4. Auerbach E
    5. Ramanna S
    6. Gunther M
    7. Glasser MF
    8. Miller KL
    9. Ugurbil K
    10. Yacoub E

    (2010) Multiplexed echo planar imaging for sub-second whole brain fMRI and fast diffusion imaging

    PLOS ONE 5:e15710.

    https://doi.org/10.1371/journal.pone.0015710

    • PubMed
    • Google Scholar
42. 1. Fields RD
    2. Bukalo O

    (2020) Myelin makes memories

    Nature Neuroscience 23:469–470.

    https://doi.org/10.1038/s41593-020-0606-x

    • Google Scholar
43. 1. Gaffan D
    2. Gaffan EA

    (1991) Amnesia in man following transection of the fornix: a review

    Brain: A Journal of Neurology 114 (Pt 6):2611–2618.

    https://doi.org/10.1093/brain/114.6.2611

    • PubMed
    • Google Scholar
44. 1. Gasser HS
    2. Grundfest H

    (1939) Axon diameters in relation to the spike dimensions and the conduction velocity in mammalian a fibers

    American Journal of Physiology-Legacy Content 127:393–414.

    https://doi.org/10.1152/ajplegacy.1939.127.2.393

    • Google Scholar
45. 1. Gilboa A
    2. Marlatte H

    (2017) Neurobiology of schemas and schema-mediated memory

    Trends in Cognitive Sciences 21:618–631.

    https://doi.org/10.1016/j.tics.2017.04.013

    • PubMed
    • Google Scholar
46. 1. Hassabis D
    2. Maguire EA

    (2007) Deconstructing episodic memory with construction

    Trends in Cognitive Sciences 11:299–306.

    https://doi.org/10.1016/j.tics.2007.05.001

    • Google Scholar
47. 1. Hodgetts CJ
    2. Postans M
    3. Warne N
    4. Varnava A
    5. Lawrence AD
    6. Graham KS

    (2017) Distinct contributions of the fornix and inferior longitudinal fasciculus to episodic and semantic autobiographical memory

    Cortex; a Journal Devoted to the Study of the Nervous System and Behavior 94:1–14.

    https://doi.org/10.1016/j.cortex.2017.05.010

    • PubMed
    • Google Scholar
48. 1. Hua K
    2. Zhang J
    3. Wakana S
    4. Jiang H
    5. Li X
    6. Reich DS
    7. Calabresi PA
    8. Pekar JJ
    9. van Zijl PCM
    10. Mori S

    (2008) Tract probability maps in stereotaxic spaces: analyses of white matter anatomy and tract-specific quantification

    NeuroImage 39:336–347.

    https://doi.org/10.1016/j.neuroimage.2007.07.053

    • PubMed
    • Google Scholar
49. 1. Hursh JB

    (1939) Conduction velocity and diameter of nerve fibers

    American Journal of Physiology-Legacy Content 127:131–139.

    https://doi.org/10.1152/ajplegacy.1939.127.1.131

    • Google Scholar
50. 1. Huxley AF
    2. Stampfli R

    (1949) Evidence for saltatory conduction in peripheral myelinated nerve fibres

    The Journal of Physiology 108:315–339.

    https://doi.org/10.1113/jphysiol.1949.sp004335

    • PubMed
    • Google Scholar
51. 1. Irish M
    2. Hornberger M
    3. El Wahsh S
    4. Lam BYK
    5. Lah S
    6. Miller L
    7. Hsieh S
    8. Hodges JR
    9. Piguet O

    (2014) Grey and white matter correlates of recent and remote autobiographical memory retrieval-insights from the dementias

    PLOS ONE 9:e113081.

    https://doi.org/10.1371/journal.pone.0113081

    • PubMed
    • Google Scholar
52. 1. Jensen JH
    2. Helpern JA

    (2010) MRI quantification of non-Gaussian water diffusion by kurtosis analysis

    NMR in Biomedicine 23:698–710.

    https://doi.org/10.1002/nbm.1518

    • PubMed
    • Google Scholar
53. 1. Jones DK
    2. Knösche TR
    3. Turner R

    (2013a) White matter integrity, fiber count, and other fallacies: the do’s and don’ts of diffusion MRI

    NeuroImage 73:239–254.

    https://doi.org/10.1016/j.neuroimage.2012.06.081

    • PubMed
    • Google Scholar
54. 1. Jones DK
    2. Christiansen KF
    3. Chapman RJ
    4. Aggleton JP

    (2013b) Distinct subdivisions of the cingulum bundle revealed by diffusion MRI fibre tracking

    Implications for Neuropsychological Investigations Neuropsychologia 51:67–78.

    https://doi.org/10.1016/j.neuropsychologia.2012.11.018

    • Google Scholar
55. 1. Jones DK
    2. Alexander DC
    3. Bowtell R
    4. Cercignani M
    5. Dell’Acqua F
    6. McHugh DJ
    7. Miller KL
    8. Palombo M
    9. Parker GJM
    10. Rudrapatna US
    11. Tax CMW

    (2018) Microstructural imaging of the human brain with a “super-scanner”: 10 key advantages of ultra-strong gradients for diffusion MRI

    NeuroImage 182:8–38.

    https://doi.org/10.1016/j.neuroimage.2018.05.047

    • PubMed
    • Google Scholar
56. 1. Jung W
    2. Lee J
    3. Shin HG
    4. Nam Y
    5. Zhang H
    6. Oh SH
    7. Lee J

    (2018) Whole brain g-ratio mapping using myelin water imaging (MWI) and neurite orientation dispersion and density imaging (NODDI)

    NeuroImage 182:379–388.

    https://doi.org/10.1016/j.neuroimage.2017.09.053

    • PubMed
    • Google Scholar
57. 1. Kaller MS
    2. Lazari A
    3. Blanco-Duque C
    4. Sampaio-Baptista C
    5. Johansen-Berg H

    (2017) Myelin plasticity and behaviour-connecting the dots

    Current Opinion in Neurobiology 47:86–92.

    https://doi.org/10.1016/j.conb.2017.09.014

    • PubMed
    • Google Scholar
58. 1. Kapur N
    2. Young A
    3. Bateman D
    4. Kennedy P

    (1989) Focal retrograde amnesia: a long term clinical and neuropsychological follow-up

    Cortex; a Journal Devoted to the Study of the Nervous System and Behavior 25:387–402.

    https://doi.org/10.1016/s0010-9452(89)80053-x

    • PubMed
    • Google Scholar
59. 1. Lakhani B
    2. Borich MR
    3. Jackson JN
    4. Wadden KP
    5. Peters S
    6. Villamayor A
    7. MacKay AL
    8. Vavasour IM
    9. Rauscher A
    10. Boyd LA

    (2016) Motor skill acquisition promotes human brain myelin plasticity

    Neural Plasticity 2016:7526135.

    https://doi.org/10.1155/2016/7526135

    • PubMed
    • Google Scholar
60. 1. LePort AKR
    2. Mattfeld AT
    3. Dickinson-Anson H
    4. Fallon JH
    5. Stark CEL
    6. Kruggel F
    7. Cahill L
    8. McGaugh JL

    (2012) Behavioral and neuroanatomical investigation of highly superior autobiographical memory (HSAM)

    Neurobiology of Learning and Memory 98:78–92.

    https://doi.org/10.1016/j.nlm.2012.05.002

    • PubMed
    • Google Scholar
61. 1. Levine B
    2. Svoboda E
    3. Hay JF
    4. Winocur G
    5. Moscovitch M

    (2002) Aging and autobiographical memory: dissociating episodic from semantic retrieval

    Psychology and Aging 17:677–689.

    https://doi.org/10.1037/0882-7974.17.4.677

    • PubMed
    • Google Scholar
62. 1. Levine B
    2. Svoboda E
    3. Turner GR
    4. Mandic M
    5. Mackey A

    (2009) Behavioral and functional neuroanatomical correlates of anterograde autobiographical memory in isolated retrograde amnesic patient M.L

    Neuropsychologia 47:2188–2196.

    https://doi.org/10.1016/j.neuropsychologia.2008.12.026

    • PubMed
    • Google Scholar
63. 1. Lorio S
    2. Fresard S
    3. Adaszewski S
    4. Kherif F
    5. Chowdhury R
    6. Frackowiak RS
    7. Ashburner J
    8. Helms G
    9. Weiskopf N
    10. Lutti A
    11. Draganski B

    (2016) New tissue priors for improved automated classification of subcortical brain structures on MRI

    NeuroImage 130:157–166.

    https://doi.org/10.1016/j.neuroimage.2016.01.062

    • PubMed
    • Google Scholar
64. 1. Lutti A
    2. Hutton C
    3. Finsterbusch J
    4. Helms G
    5. Weiskopf N

    (2010) Optimization and validation of methods for mapping of the radiofrequency transmit field at 3T

    Magnetic Resonance in Medicine 64:229–238.

    https://doi.org/10.1002/mrm.22421

    • PubMed
    • Google Scholar
65. 1. Macdonald J
    2. Ruthotto L

    (2018) Improved susceptibility artifact correction of echo-planar MRI using the alternating direction method of multipliers

    Journal of Mathematical Imaging and Vision 60:268–282.

    https://doi.org/10.1007/s10851-017-0757-x

    • Google Scholar
66. 1. Maguire EA
    2. Gadian DG
    3. Johnsrude IS
    4. Good CD
    5. Ashburner J
    6. Frackowiak RSJ
    7. Frith CD

    (2000) Navigation-related structural change in the hippocampi of taxi drivers

    PNAS 97:4398–4403.

    https://doi.org/10.1073/pnas.070039597

    • PubMed
    • Google Scholar
67. 1. Maguire EA

    (2001) Neuroimaging studies of autobiographical event memory

    Philosophical Transactions of the Royal Society of London Series B: Biological Sciences 356:1441–1451.

    https://doi.org/10.1098/rstb.2001.0944

    • Google Scholar
68. 1. Maguire EA
    2. Valentine ER
    3. Wilding JM
    4. Kapur N

    (2003) Routes to remembering: the brains behind superior memory

    Nature Neuroscience 6:90–95.

    https://doi.org/10.1038/nn988

    • Google Scholar
69. 1. Maguire EA

    (2012) Studying the freely-behaving brain with fMRI

    NeuroImage 62:1170–1176.

    https://doi.org/10.1016/j.neuroimage.2012.01.009

    • Google Scholar
70. 1. Marek S
    2. Tervo-Clemmens B
    3. Calabro FJ
    4. Montez DF
    5. Kay BP
    6. Hatoum AS
    7. Donohue MR
    8. Foran W
    9. Miller RL
    10. Hendrickson TJ
    11. Malone SM
    12. Kandala S
    13. Feczko E
    14. Miranda-Dominguez O
    15. Graham AM
    16. Earl EA
    17. Perrone AJ
    18. Cordova M
    19. Doyle O
    20. Moore LA
    21. Conan GM
    22. Uriarte J
    23. Snider K
    24. Lynch BJ
    25. Wilgenbusch JC
    26. Pengo T
    27. Tam A
    28. Chen J
    29. Newbold DJ
    30. Zheng A
    31. Seider NA
    32. Van AN
    33. Metoki A
    34. Chauvin RJ
    35. Laumann TO
    36. Greene DJ
    37. Petersen SE
    38. Garavan H
    39. Thompson WK
    40. Nichols TE
    41. Yeo BTT
    42. Barch DM
    43. Luna B
    44. Fair DA
    45. Dosenbach NUF

    (2022) Reproducible brain-wide association studies require thousands of individuals

    Nature 603:654–660.

    https://doi.org/10.1038/s41586-022-04492-9

    • Google Scholar
71. 1. McCormick C
    2. Ciaramelli E
    3. De Luca F
    4. Maguire EA

    (2018) Comparing and contrasting the cognitive effects of hippocampal and ventromedial prefrontal cortex damage: a review of human lesion studies

    Neuroscience 374:295–318.

    https://doi.org/10.1016/j.neuroscience.2017.07.066

    • PubMed
    • Google Scholar
72. 1. McCormick C
    2. Barry DN
    3. Jafarian A
    4. Barnes GR
    5. Maguire EA

    (2020) vmPFC drives hippocampal processing during autobiographical memory recall regardless of remoteness

    Cerebral Cortex 30:5972–5987.

    https://doi.org/10.1093/cercor/bhaa172

    • PubMed
    • Google Scholar
73. 1. McDermott KB
    2. Szpunar KK
    3. Christ SE

    (2009) Laboratory-Based and autobiographical retrieval tasks differ substantially in their neural substrates

    Neuropsychologia 47:2290–2298.

    https://doi.org/10.1016/j.neuropsychologia.2008.12.025

    • PubMed
    • Google Scholar
74. 1. Memel M
    2. Wank AA
    3. Ryan L
    4. Grilli MD

    (2020) The relationship between episodic detail generation and anterotemporal, posteromedial, and hippocampal white matter tracts

    Cortex; a Journal Devoted to the Study of the Nervous System and Behavior 123:124–140.

    https://doi.org/10.1016/j.cortex.2019.10.010

    • PubMed
    • Google Scholar
75. 1. Meng X
    2. Rosenthal R
    3. Rubin DB

    (1992) Comparing correlated correlation coefficients

    Psychological Bulletin 111:172–175.

    https://doi.org/10.1037/0033-2909.111.1.172

    • Google Scholar
76. 1. Miller EM

    (1994) Intelligence and brain myelination: a hypothesis

    Personality and Individual Differences 17:803–832.

    https://doi.org/10.1016/0191-8869(94)90049-3

    • Google Scholar
77. 1. Miller CT
    2. Gire D
    3. Hoke K
    4. Huk AC
    5. Kelley D
    6. Leopold DA
    7. Smear MC
    8. Theunissen F
    9. Yartsev M
    10. Niell CM

    (2022) Natural behavior is the language of the brain

    Current Biology 32:R482–R493.

    https://doi.org/10.1016/j.cub.2022.03.031

    • PubMed
    • Google Scholar
78. 1. Mobbs D
    2. Wise T
    3. Suthana N
    4. Guzmán N
    5. Kriegeskorte N
    6. Leibo JZ

    (2021) Promises and challenges of human computational ethology

    Neuron 109:2224–2238.

    https://doi.org/10.1016/j.neuron.2021.05.021

    • PubMed
    • Google Scholar
79. 1. Mohammadi S
    2. Freund P
    3. Feiweier T
    4. Curt A
    5. Weiskopf N

    (2013) The impact of post-processing on spinal cord diffusion tensor imaging

    NeuroImage 70:377–385.

    https://doi.org/10.1016/j.neuroimage.2012.12.058

    • PubMed
    • Google Scholar
80. 1. Mohammadi S
    2. Carey D
    3. Dick F
    4. Diedrichsen J
    5. Sereno MI
    6. Reisert M
    7. Callaghan MF
    8. Weiskopf N

    (2015) Whole-Brain in-vivo measurements of the axonal g-ratio in a group of 37 healthy volunteers

    Frontiers in Neuroscience 9:441.

    https://doi.org/10.3389/fnins.2015.00441

    • PubMed
    • Google Scholar
81. 1. Mohammadi S
    2. Callaghan MF

    (2021) Towards in vivo g-ratio mapping using MRI: unifying myelin and diffusion imaging

    Journal of Neuroscience Methods 348:108990.

    https://doi.org/10.1016/j.jneumeth.2020.108990

    • PubMed
    • Google Scholar
82. 1. Mori S
    2. Oishi K
    3. Jiang H
    4. Jiang L
    5. Li X
    6. Akhter K
    7. Hua K
    8. Faria AV
    9. Mahmood A
    10. Woods R
    11. Toga AW
    12. Pike GB
    13. Neto PR
    14. Evans A
    15. Zhang J
    16. Huang H
    17. Miller MI
    18. Zijl P
    19. Mazziotta J

    (2008) Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template

    NeuroImage 40:570–582.

    https://doi.org/10.1016/j.neuroimage.2007.12.035

    • Google Scholar
83. 1. Movahedian Attar F
    2. Kirilina E
    3. Haenelt D
    4. Pine KJ
    5. Trampel R
    6. Edwards LJ
    7. Weiskopf N

    (2020) Mapping short association fibers in the early cortical visual processing stream using in vivo diffusion tractography

    Cerebral Cortex 30:4496–4514.

    https://doi.org/10.1093/cercor/bhaa049

    • Google Scholar
84. 1. Mullally SL
    2. Maguire EA

    (2011) A new role for the parahippocampal cortex in representing space

    The Journal of Neuroscience 31:7441–7449.

    https://doi.org/10.1523/JNEUROSCI.0267-11.2011

    • PubMed
    • Google Scholar
85. 1. Nastase SA
    2. Goldstein A
    3. Hasson U

    (2020) Keep it real: rethinking the primacy of experimental control in cognitive neuroscience

    NeuroImage 222:117254.

    https://doi.org/10.1016/j.neuroimage.2020.117254

    • Google Scholar
86. 1. Nave KA

    (2010) Myelination and support of axonal integrity by glia

    Nature 468:244–252.

    https://doi.org/10.1038/nature09614

    • PubMed
    • Google Scholar
87. Conference

    1. Oeschger J
    2. Tabelow K
    3. Mohammadi S

    (2021)

    Reducing rician noise bias in axial-symmetric diffusion kurtosis imaging and biophysical tissue models

    Proceedings of the International Society for Magnetic Resonance in Medicine Scientific Meeting and Exhibition.

    • Google Scholar
88. 1. Palombo DJ
    2. Alain C
    3. Söderlund H
    4. Khuu W
    5. Levine B

    (2015) Severely deficient autobiographical memory (SDAM) in healthy adults: a new mnemonic syndrome

    Neuropsychologia 72:105–118.

    https://doi.org/10.1016/j.neuropsychologia.2015.04.012

    • PubMed
    • Google Scholar
89. 1. Palombo DJ
    2. Sheldon S
    3. Levine B

    (2018) Individual differences in autobiographical memory

    Trends in Cognitive Sciences 22:583–597.

    https://doi.org/10.1016/j.tics.2018.04.007

    • PubMed
    • Google Scholar
90. 1. Papagno C
    2. Miracapillo C
    3. Casarotti A
    4. Romero Lauro LJ
    5. Castellano A
    6. Falini A
    7. Casaceli G
    8. Fava E
    9. Bello L

    (2011) What is the role of the uncinate fasciculus? Surgical removal and proper name retrieval

    Brain 134:405–414.

    https://doi.org/10.1093/brain/awq283

    • PubMed
    • Google Scholar
91. 1. Peters A
    2. Moss MB
    3. Sethares C

    (2000) Effects of aging on myelinated nerve fibers in monkey primary visual cortex

    The Journal of Comparative Neurology 419:364–376.

    https://doi.org/10.1002/(sici)1096-9861(20000410)419:3<364::aid-cne8>3.0.co;2-r

    • PubMed
    • Google Scholar
92. 1. Peters A
    2. Sethares C

    (2002) Aging and the myelinated fibers in prefrontal cortex and corpus callosum of the monkey

    The Journal of Comparative Neurology 442:277–291.

    https://doi.org/10.1002/cne.10099

    • PubMed
    • Google Scholar
93. 1. Reed TE
    2. Jensen AR

    (1992) Conduction velocity in a brain nerve pathway of normal adults correlates with intelligence level

    Intelligence 16:259–272.

    https://doi.org/10.1016/0160-2896(92)90009-G

    • Google Scholar
94. 1. Reed ET
    2. Jensen AR

    (1993) A somatosensory latency between the thalamus and cortex also correlates with level of intelligence

    Intelligence 17:443–450.

    https://doi.org/10.1016/0160-2896(93)90011-S

    • Google Scholar
95. 1. Rey A

    (1941)

    L'examen psychologique dans les cas d’encéphalopathie traumatique

    Archives de Psychologie 28:215–285.

    • Google Scholar
96. 1. Rushton WAH

    (1951) A theory of the effects of fibre size in medullated nerve

    The Journal of Physiology 115:101–122.

    https://doi.org/10.1113/jphysiol.1951.sp004655

    • PubMed
    • Google Scholar
97. 1. Ruthotto L
    2. Kugel H
    3. Olesch J
    4. Fischer B
    5. Modersitzki J
    6. Burger M
    7. Wolters CH

    (2012) Diffeomorphic susceptibility artifact correction of diffusion-weighted magnetic resonance images

    Physics in Medicine and Biology 57:5715–5731.

    https://doi.org/10.1088/0031-9155/57/18/5715

    • PubMed
    • Google Scholar
98. 1. Salami M
    2. Itami C
    3. Tsumoto T
    4. Kimura F

    (2003) Change of conduction velocity by regional myelination yields constant latency irrespective of distance between thalamus and cortex

    PNAS 100:6174–6179.

    https://doi.org/10.1073/pnas.0937380100

    • PubMed
    • Google Scholar
99. 1. Schacter DL
    2. Addis DR
    3. Hassabis D
    4. Martin VC
    5. Spreng RN
    6. Szpunar KK

    (2012) The future of memory: remembering, imagining, and the brain

    Neuron 76:677–694.

    https://doi.org/10.1016/j.neuron.2012.11.001

    • PubMed
    • Google Scholar
100. 1. Schmidt H
     2. Knösche TR

     (2019) Action potential propagation and synchronisation in myelinated axons

     PLOS Computational Biology 15:e1007004.

     https://doi.org/10.1371/journal.pcbi.1007004

     • Google Scholar
101. 1. Scoville WB
     2. Milner B

     (1957) Loss of recent memory after bilateral hippocampal lesions

     Journal of Neurology, Neurosurgery, and Psychiatry 20:11–21.

     https://doi.org/10.1136/jnnp.20.1.11

     • PubMed
     • Google Scholar
102. 1. Shastin D
     2. Genc S
     3. Parker GD
     4. Koller K
     5. Tax CMW
     6. Evans J
     7. Hamandi K
     8. Gray WP
     9. Jones DK
     10. Chamberland M

     (2022) Surface-based tracking for short association fibre tractography

     NeuroImage 260:119423.

     https://doi.org/10.1016/j.neuroimage.2022.119423

     • PubMed
     • Google Scholar
103. 1. Spiers HJ
     2. Maguire EA

     (2007) Decoding human brain activity during real-world experiences

     Trends in Cognitive Sciences 11:356–365.

     https://doi.org/10.1016/j.tics.2007.06.002

     • Google Scholar
104. 1. Spreng RN
     2. Mar RA
     3. Kim ASN

     (2009) The common neural basis of autobiographical memory, prospection, navigation, theory of mind, and the default mode: a quantitative meta-analysis

     Journal of Cognitive Neuroscience 21:489–510.

     https://doi.org/10.1162/jocn.2008.21029

     • PubMed
     • Google Scholar
105. 1. Steriade M

     (1995) Two channels in the cerebellothalamocortical system

     The Journal of Comparative Neurology 354:57–70.

     https://doi.org/10.1002/cne.903540106

     • PubMed
     • Google Scholar
106. 1. Stikov N
     2. Campbell JSW
     3. Stroh T
     4. Lavelée M
     5. Frey S
     6. Novek J
     7. Nuara S
     8. Ho MK
     9. Bedell BJ
     10. Dougherty RF
     11. Leppert IR
     12. Boudreau M
     13. Narayanan S
     14. Duval T
     15. Cohen-Adad J
     16. Picard PA
     17. Gasecka A
     18. Côté D
     19. Pike GB

     (2015) In vivo histology of the myelin g-ratio with magnetic resonance imaging

     NeuroImage 118:397–405.

     https://doi.org/10.1016/j.neuroimage.2015.05.023

     • PubMed
     • Google Scholar
107. Book

     1. Strauss E
     2. Sherman EM
     3. Spreen O

     (2006)

     A Compendium of Neuropsychological Tests: Administration, Commentary and Norms

     Oxford University Press.

     • Google Scholar
108. 1. Svoboda E
     2. McKinnon MC
     3. Levine B

     (2006) The functional neuroanatomy of autobiographical memory: a meta-analysis

     Neuropsychologia 44:2189–2208.

     https://doi.org/10.1016/j.neuropsychologia.2006.05.023

     • PubMed
     • Google Scholar
109. 1. Tabelow K
     2. Balteau E
     3. Ashburner J
     4. Callaghan MF
     5. Draganski B
     6. Helms G
     7. Kherif F
     8. Leutritz T
     9. Lutti A
     10. Phillips C
     11. Reimer E
     12. Ruthotto L
     13. Seif M
     14. Weiskopf N
     15. Ziegler G
     16. Mohammadi S

     (2019) HMRI-a toolbox for quantitative MRI in neuroscience and clinical research

     NeuroImage 194:191–210.

     https://doi.org/10.1016/j.neuroimage.2019.01.029

     • PubMed
     • Google Scholar
110. 1. Tsivilis D
     2. Vann SD
     3. Denby C
     4. Roberts N
     5. Mayes AR
     6. Montaldi D
     7. Aggleton JP

     (2008) A disproportionate role for the fornix and mammillary bodies in recall versus recognition memory

     Nature Neuroscience 11:834–842.

     https://doi.org/10.1038/nn.2149

     • PubMed
     • Google Scholar
111. 1. Tulving E

     (2002) Episodic memory: from mind to brain

     Annual Review of Psychology 53:1–25.

     https://doi.org/10.1146/annurev.psych.53.100901.135114

     • PubMed
     • Google Scholar
112. 1. Vann SD
     2. Aggleton JP
     3. Maguire EA

     (2009) What does the retrosplenial cortex do?

     Nature Reviews Neuroscience 10:792–802.

     https://doi.org/10.1038/nrn2733

     • PubMed
     • Google Scholar
113. 1. Van Petten C

     (2004) Relationship between hippocampal volume and memory ability in healthy individuals across the lifespan: review and meta-analysis

     Neuropsychologia 42:1394–1413.

     https://doi.org/10.1016/j.neuropsychologia.2004.04.006

     • PubMed
     • Google Scholar
114. 1. Veraart J
     2. Nunes D
     3. Rudrapatna U
     4. Fieremans E
     5. Jones DK
     6. Novikov DS
     7. Shemesh N

     (2020) Noninvasive quantification of axon radii using diffusion MRI

     eLife 9:e49855.

     https://doi.org/10.7554/eLife.49855

     • PubMed
     • Google Scholar
115. 1. Von Der Heide RJ
     2. Skipper LM
     3. Klobusicky E
     4. Olson IR

     (2013) Dissecting the uncinate fasciculus: disorders, controversies and a hypothesis

     Brain 136:1692–1707.

     https://doi.org/10.1093/brain/awt094

     • PubMed
     • Google Scholar
116. Book

     1. Warrington EK

     (1984)

     Recognition Memory Test: Manual

     NFER-Nelson.

     • Google Scholar
117. 1. Waxman SG

     (1980) Determinants of conduction velocity in myelinated nerve fibers

     Muscle & Nerve 3:141–150.

     https://doi.org/10.1002/mus.880030207

     • PubMed
     • Google Scholar
118. Book

     1. Wechsler D

     (2009)

     WMS-IV.: Wechsler Memory Scale-Administration and Scoring Manual

     Pearson Assessment.

     • Google Scholar
119. 1. Weiskopf N
     2. Suckling J
     3. Williams G
     4. Correia MM
     5. Inkster B
     6. Tait R
     7. Ooi C
     8. Bullmore ET
     9. Lutti A

     (2013) Quantitative multi-parameter mapping of R1, PD (*), MT, and R2 (*) at 3T: a multi-center validation

     Frontiers in Neuroscience 7:95.

     https://doi.org/10.3389/fnins.2013.00095

     • PubMed
     • Google Scholar
120. 1. West KL
     2. Kelm ND
     3. Carson RP
     4. Does MD

     (2016) A revised model for estimating g-ratio from MRI

     NeuroImage 125:1155–1158.

     https://doi.org/10.1016/j.neuroimage.2015.08.017

     • Google Scholar
121. 1. West KL
     2. Kelm ND
     3. Carson RP
     4. Alexander DC
     5. Gochberg DF
     6. Does MD

     (2018) Experimental studies of g-ratio MRI in ex vivo mouse brain

     NeuroImage 167:366–371.

     https://doi.org/10.1016/j.neuroimage.2017.11.064

     • Google Scholar
122. 1. Winocur G
     2. Moscovitch M

     (2011) Memory transformation and systems consolidation

     Journal of the International Neuropsychological Society 17:766–780.

     https://doi.org/10.1017/S1355617711000683

     • PubMed
     • Google Scholar
123. 1. Woollett K
     2. Maguire EA

     (2011) Acquiring “the knowledge” of London’s layout drives structural brain changes

     Current Biology 21:2109–2114.

     https://doi.org/10.1016/j.cub.2011.11.018

     • Google Scholar
124. 1. Xi MC
     2. Liu RH
     3. Engelhardt JK
     4. Morales FR
     5. Chase MH

     (1999) Changes in the axonal conduction velocity of pyramidal tract neurons in the aged cat

     Neuroscience 92:219–225.

     https://doi.org/10.1016/S0306-4522(98)00754-4

     • Google Scholar
125. 1. Xin W
     2. Chan JR

     (2020) Myelin plasticity: sculpting circuits in learning and memory

     Nature Reviews. Neuroscience 21:682–694.

     https://doi.org/10.1038/s41583-020-00379-8

     • PubMed
     • Google Scholar
126. 1. Xu J
     2. Moeller S
     3. Auerbach EJ
     4. Strupp J
     5. Smith SM
     6. Feinberg DA
     7. Yacoub E
     8. Uğurbil K

     (2013) Evaluation of slice accelerations using multiband echo planar imaging at 3T

     NeuroImage 83:991–1001.

     https://doi.org/10.1016/j.neuroimage.2013.07.055

     • Google Scholar
127. 1. Yu F
     2. Fan Q
     3. Tian Q
     4. Ngamsombat C
     5. Machado N
     6. Bireley JD
     7. Russo AW
     8. Nummenmaa A
     9. Witzel T
     10. Wald LL
     11. Klawiter EC
     12. Huang SY

     (2019) Imaging g-ratio in multiple sclerosis using high-gradient diffusion

     MRI and Macromolecular Tissue Volume American Journal of Neuroradiology 40:1871.

     https://doi.org/10.3174/ajnr.A6283

     • Google Scholar
128. 1. Zhang H
     2. Schneider T
     3. Wheeler-Kingshott CA
     4. Alexander DC

     (2012) NODDI: practical in vivo neurite orientation dispersion and density imaging of the human brain

     NeuroImage 61:1000–1016.

     https://doi.org/10.1016/j.neuroimage.2012.03.072

     • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "Signalling speed along a key white matter tract influences memory recall ability in humans" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Laura Colgin as the Senior Editor. The reviewers have opted to remain anonymous.

We are sorry to say that, after consultation with the reviewers, we have decided that the work as described in the current manuscript will not be considered for publication by eLife. However, eLife might consider a very substantial revision, if the authors are interested in undertaking it. The new version would need to address all of the reviewers' concerns, including incorporating reviewer 1's suggestion to report correlations between white matter tracts and all cognitive measures collected, addressing concerns about the g-ratio measurement, as well as changing claims about measuring signaling speed and removing language suggesting causal relationships.

Reviewer #1 (Recommendations for the authors):

My main recommendation is that the authors report the raw correlations between other memory tests and the MR g-ratios of their candidate white matter tracts (assuming that this is possible). This will either show that increased g-ratios in the parahippocampal cingulum bundle are a general feature of individuals who perform better on memory tests, or it is something more specific to the scores on the Autobiographical Memory Interview. It might be the case that other tracts show associations with other tests. Any of these scenarios would help contextualize the results of the paper as they stand.

I am not particularly concerned about correcting for multiple comparisons or arbitrary thresholds for significance. I think that the study is important, novel, and well carried out and that the findings will be important to researchers in several fields. If it is the case that reliable associations between cognitive measures and the MR g-ratio cannot be established in a sample of 217 people then this is really useful information!

I have probably misunderstood the interpretation of the findings with respect to the examples shown in figure 1. It seemed that the authors interpreted their positive correlation between internal details and the g-ratio in the parahippocampal cingulum bundle as evidence for option D – people with better memory recall scores are likely to have larger inner axon diameters and relatively smaller increases in myelin. However, since example 1E involves no change in the g-ratio, I couldn't understand why the finding of no relationship between memory recall ability and g-ratio from the fornix or uncinate fasciculus was not taken as support for option E. It might help if the authors could clarify which situation(s) would lead to them accepting option E as the best explanation of their results.

The following paper by Hodgetts et al., seems very relevant – the present study fails to replicate their findings related to the fornix.

Hodgetts, C. J., Postans, M., Warne, N., Varnava, A., Lawrence, A. D., and Graham, K. S. (2017). Distinct contributions of the fornix and inferior longitudinal fasciculus to episodic and semantic autobiographical memory. cortex, 94, 1-14.

Reviewer #2 (Recommendations for the authors):

– Following what I mentioned in the public review, given the lack of any other measure to quantitatively or even qualitatively assess either conduction velocity/delays or axon size, the authors should refrain from directly claiming that larger axons mean better memory recall abilities, which instead is done in the abstract (page 2, "(…) two tract features seemed to favour better memory retrieval – large inner axon diameters (…)" – despite the axon diameters not being measured or estimated), and in the discussion (page 18, "(…) we were able to identify two particular features (…) that might favour better memory recall, namely larger inner axon diameters (…)").

– Related to the previous point, a suggestion that I have is to include, among the exhaustive comparisons performed in the appendix, the ones with the MTsat values. As an absolute measure of myelin, MTsat could further help the authors by providing a complete picture of their results.

– It is rather peculiar that the abstract does not mention at all myelin (which in the end is the main protagonist of the study) and that does not briefly define the g-ratio. I understand the authors' focus on the conduction velocity, but these additions would surely benefit a more general audience.

– Related to the previous point, the authors may also consider rethinking the use of "signalling speed", which basically appears only in the title and in the abstract, in favour of "conduction velocity" which is used extensively in the text.

– On page 3, the authors stated: "The scope for signals to travel quickly and efficiently between brain regions could make all the differences for an individual's survival". Although the idea that faster conduction velocity constitutes an evolutionary advantage could be appealing and the peripheral nervous system (PNS) supports the concept extremely well (a very famous example is depicted in Wedel, Acta Palaeontologica 2011), there are several reasons to be cautious with such a statement. In contrast to what happens in the PNS (where faster propagation reflects in faster sensory processing and faster motor responses), in the central nervous system conduction phenomena serve the purpose of guaranteeing synchrony in the activity of different neuronal populations (Seidtl, Neuroscience 2014; Gollo et al., NeuroImage 2017; Pariz et al., Plos Computational Biology 2021) and the role of delays could be more subtle. Unless the authors can integrate the statement with some references to experimental evidence, I would remove such a statement.

– Still on page 3, the authors mentioned that the conduction velocity "is dependent upon two biological features – the axonal diameter and the presence and thickness of a myelin sheath". This picture is too approximative as experimental work (Arancibia-Carcamo et al., eLife 2017) and simulations (Drakesmith et al., NeuroImage 2019) have shown how several parameters influence conduction velocity, potentially even with a larger impact than the g-ratio. It would be appropriate for a more general audience to add more details. Similarly, on page 4 it would help to refer to saltatory propagation when explaining the effect of myelin as an insulator.

– On page 7, there is a bit of a leap between the section on the uses and potentials of the MRI-based g-ratio and the proposed application in autobiographical memory. It would help to provide a couple of sentences to link those together or to rearrange a bit the paragraph in-between pages 7 and 8, where at the very end the case for g-ratio and memory is made.

– The considerations in figure 1 made on pages 4-5 are at the moment very hard without referring to an actual relationship between conduction velocity, axonal diameter, and g-ratio. Even though the authors are not actually computing conduction velocity, it would help to refer to established formulas such as the one from Rushton (Rushton, Journal of Physiology 1951) or the one form Waxman and Bennet (Waxman and Bennet, Nature New Biology 1972).

– I noticed that in Figures 4 and 5 the scatterplots are based on the residuals rather than the actual g-ratio and neurite dispersion values. Can I ask what is the reason behind this choice? This representation makes actually harder to get a visual sense of the dynamics of the reported relationships.

– On page 18, the authors mentioned how these results can provide a new perspective on the underlying neural substrate of autographical memory, and stated: "This is especially welcome given the dearth of consistent findings linking hippocampal grey matter volume or microstructure with autobiographical memory recall ability in the healthy population". Can the authors expand their argument here? I am missing the link between the current findings (on white matter microstructure) with previous ones on grey matter features.

– On page 22, there is a reference to the CONNECTOM MRI scanners (misspelled as "connectome MRI scanners") – as the general audience is not familiar with those it would help to add a brief sentence explaining what they are and maybe a reference (for example, Jones et al., NeuroImage 2018).

– Can the authors comment on the advantages of the proposed ROI-based approach compared with a tract-based approach?

– I may have missed it, but I am not sure of what is the reason for the fornix tract being "not probabilistic" (page 29) – are not all the tract ROIs obtained from the John Hopkins probabilistic atlas?

– I could not find any mention of the main methodological limitations (e.g., the use of signal fractions rather than volume fractions, the bias towards larger axons, and the need for single-point calibration). There is only a brief mention in the appendix on page 43. I think these limitations should be part of the discussion to put these results in the right context.

Reviewer #3 (Recommendations for the authors):

I have some suggestions the authors may consider, presented in chronological order.

Title

The title suggests a causal relationship, while this is based on an association. Maybe better to use a neutral word, i.e. replace "influences" with e.g. "is associated with" or something similar?

As I understand MR g-ratio, it reflects structural aspects of nerve fibers, and although probably related to conduction velocity/ signaling speed, it is not a direct measure of this. Hence, I believe the title is overstating this relationship by referring to "signalling speed" directly. I fully understand the need for short and focused titles, but I believe this is a bit misleading. The same point applies to a couple of other places in the manuscript too. Of course fine to discuss this as an interpretation.

Abstract

Not all readers are probably familiar with the MR g-ratio, so maybe it can be defined in the abstract?

Since this is a cross-sectional study, I suggest avoiding phrases that imply change, e.g. "increased MR g-ratio". Maybe replace it with something neutral, such as "higher MR g-ratio"? The same is the case in several places in the manuscript, e.g. ln 195.

As with the title, the results reflect associations, not causation. This does not render them less interesting, but I would recommend replacing "favour better memory …" with e.g. "is associated with better memory …". I also wonder whether "offer a new perspective on drivers of individual differences in memory recall ability" is overstating a bit, since (a) we don't know whether g-ratio (or signal speed) is driving anything, and (b) this is a rather well-established theory (as also explained in the introduction, and it is stated ln 44 "it is perhaps no surprise that speed may be of the essence")?

Introduction

ln 51-54: Why was aging discussed? Reading the introductory paragraph, I expected aging to be part of the study, which it was not.

Ln 123-125: "While some healthy individuals can recollect decades-old autobiographical memories with great richness and clarity, others struggle to recall what they did last weekend" Yes, but as the authors surely know, this reflects different memory systems or processes, and therefore can happen within the same individual too, not only as a feature of inter-individual differences.

Ln 195-200: Figure 1 clearly illustrates different scenarios, which is very useful. However, the hypothesis that better autobiographical memory is related to more myelin could also be consistent with no relationship with the MR g-ratio if the axon diameter is larger in participants with a good memory, right? And vice versa for the hypothesis about axon diameter? I don't see how the ratio measure can distinguish between the different accounts since the measures per definition are related to two possibly independent properties which both can be of importance? The authors multiple times mention that other relevant metrics, such as those derived from DTI, lack biological specificity. But the MR g-ratio measure seems also to be unspecific in this regard. If axon diameter and myelin thickness can be estimated separately, why not use them directly in the analyses instead of the MR g-ratio? There may be a methodological reason for this, but I think this could be more clearly explained as MR g-ratio is not often reported in the literature and so will be novel to many readers. Relatedly, although there are methods papers on this topic, maybe a brief justification of how can ratios between these small compartments on the micrometer scale can be calculated from voxels on the millimeter scale already in the introduction?

Results

I understand that the authors don't want to provide the full description of results for FA and MD in the main text, but it would be convenient to report a brief summary of the main results also in the text. As far as I can see from Appendix 1, none even approach significance for fornix and uncinated fasciculus, so maybe at least state that?

This may be a matter of taste, but I would suggest moving the section on why the parahippocampal cingulum bundle should be important for autobiographical memory from the Results section (ln 236 and onwards) to the introduction or the discussion.

Ln 279: A positive relationship between MR g-ratio and memory is interpreted as reflecting larger inner axon diameters. Ref Figure 1, this can also be due to thinner myelin sheets, right, as it is a ratio measure? Multiple comparison correction was mentioned for some analyses above, was this done here two, and if so, how? Would be useful to state explicitly.

Inspecting the Appendix, it seems that the relationships between autobiographic memory and PCB FA and MD were also significant, with r-values that were at least as large as for MR g-ratio. Why not include this in the main text? These are metrics of interest to many researchers. You could also run a multiple regression analysis, and test whether each metric gives unique information or whether they are redundant. I think this would be a very useful addition to the manuscript, regardless of the outcome. If MR g-ratio is still significantly related to autobiographical memory, this would strengthen the main argument in the paper. If not, we will know more about the interrelationships between these different tract metrics, which also will be very useful.

Discussion

Ln 336: Is it correct to say that the MR g-ratio – memory correlation represents "neural instantiation of autobiographical memory"? To me, it seems to overstate the results, as this is an association, and we do not know whether there is any causality involved. The same goes for the last part of the sentence "drivers of individual differences in recall ability". I think the results are of sufficient interest, without such statements.

Methods, Participants

Ln 456-457: "People with hobbies or vocations known to be associated with the hippocampus (e.g. licenced London taxi drivers) were excluded." Could you please be a little more specific, most activities will be associated with the hippocampus? How many were excluded based on this?

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

The authors put an amazing effort into taking into account the reviewers' comments and into improving the paper with further details and analysis. The finding that magnetization transfer saturation does not correlate with the score of interest while the g-ratio instead does is quite interesting - I think that for the sake of fully characterizing the relationship what is missing at the moment is correlating the isotropic volume fraction with the recall score: since the g-ratio is computed using MTsat, Viso and Vic, double-checking that a correlation does not exist with Viso is the last step needed to show that indeed is the combination of all these measures that is meaningful rather than the single measures themselves.

Reviewer #1 (Recommendations for the authors):

My main recommendation is that the authors report the raw correlations between other memory tests and the MR g-ratios of their candidate white matter tracts (assuming that this is possible). This will either show that increased g-ratios in the parahippocampal cingulum bundle are a general feature of individuals who perform better on memory tests, or it is something more specific to the scores on the Autobiographical Memory Interview. It might be the case that other tracts show associations with other tests. Any of these scenarios would help contextualize the results of the paper as they stand.

As noted in our response to the Reviewer’s Public Review, our main interest over many years, and hence the focus of this study, is autobiographical memory recall because it directly relates to how people function in real life. We used the current, gold standard approach to measuring autobiographical memory recall ability (Levine’s Autobiographical Interview). Nevertheless, at the Reviewer’s request we have now analysed data from eight additional laboratory-based memory tests. Scores from these tests did not correlate with the MR g-ratio for any of our three white matter tracts of interest. Recall of vivid, detailed, multimodal, autobiographical memories may rely on inter-regional connectivity to a greater degree than simpler, more constrained laboratory-based memory tests.

Therefore, as well as speaking to conduction velocity, our findings now also contribute to wider discussions about real-world compared to laboratory-based memory tests. We thank the Reviewer again for making the excellent suggestion to include these additional data, analyses and discussion points, as noted below.

Introduction, p. 10: “All participants underwent the widely-used Autobiographical Interview (Levine et al., 2002). This task, which is the gold standard in the field, provides a detailed metric characterising a person’s ability to recall real-life past experiences, as well as providing a useful control measure. We also examined performance on another eight standard laboratory-based memory tests. Their inclusion allowed us to ascertain whether any relationships with conduction velocity were specific to recollecting detailed autobiographical memories from real life, or were also applicable to the recall of more constrained laboratory-based stimuli.”

Introduction, pp. 11-12: “Finally, given that a previous functional MRI meta-analysis identified different neural substrates associated with autobiographical memory recall and the recall of laboratory-based stimuli (McDermott et al., 2009; see also Maguire, 2001, 2012; Miller et al., 2022; Mobbs et al., 2021; Nastase et al., 2020; Spiers and Maguire, 2007), we expected that associations with the MR g-ratio might be specific to autobiographical memory recall. We reasoned that detailed, multimodal, autobiographical memories may rely on inter-regional connectivity to a greater degree than simpler, more constrained laboratory-based memory tests.”

Results, pp. 12-13:

“Laboratory-based memory test performance

While our main interest was in autobiographical memory recall, eight commonly used laboratory-based memory tasks were also administered. Their inclusion allowed us to establish whether any associations identified with the main microstructure measure, MR g-ratio, were specific or not to the recollection of real-life autobiographical memories.

The ability to recall a short narrative was examined using the immediate and delayed recall tests of the Logical Memory subtest of the Wechsler Memory Scale IV (Wechsler, 2009). Across participants, the mean immediate recall scaled score was 12.95 (SD = 2.09, range = 6-18) and the mean delayed recall scaled score was 12.58 (SD = 2.62, range = 6-19). Verbal list recall ability was assessed using the immediate and delayed recall of the Rey Auditory Verbal Learning Test (see Strauss et al., 2006). The mean immediate recall (aggregate) score was 58.82 (SD = 7.42, range = 33-73) and the mean delayed recall score was 12.92 (SD = 2.17, range = 6-15). Visuospatial recall ability was examined using the delayed recall of the Rey–Osterrieth Complex Figure (Rey, 1941), with a mean delayed recall score of 22.28 (SD = 5.71, range = 8.5-35). Recognition memory ability was tested using the Warrington Recognition Memory Tests for Words and Faces (Warrington, 1984). The mean recognition memory scaled score for words was 12.75 (SD = 2.06, range = 3-15) and for faces was 11.00 (SD = 3.33, range = 3-18). Finally, participants also completed the “Dead or Alive” task which probes general knowledge about whether famous people have died or are still alive, providing a measure of semantic memory (Kapur et al., 1989). The mean accuracy performance on this test was 81.32% (SD = 8.44, range = 57.14%-97.26%).”

Results, p. 14: “There were no significant associations between fornix or uncinate fasciculus MR g-ratios or magnetisation transfer saturation values and any of the laboratory-based memory tests (see Appendix 1– tables 5 and 6 for full details, source data are available in Supplementary file 1).”

Results, p. 18:

“No relationship between microstructure measures of the parahippocampal cingulum bundle and scores on laboratory-based memory tests

Finally, we tested for associations between the key microstructure measures and performance on the eight laboratory-based memory tasks. No relationships with the MR g-ratio or magnetisation transfer saturation values were evident for any task, even when using an uncorrected p < 0.05 threshold (Appendix 1-table 21, source data are available in Supplementary file 2). This suggests that the parahippocampal cingulum bundle MR g-ratio, and by inference conduction velocity, was specifically associated with recall of autobiographical memories from real life rather than performance on the more constrained laboratory-based memory tests.”

Discussion, p. 24: “Moreover, no relationships were identified with any of the eight laboratory-based memory tasks examined, highlighting the specific nature of the relationship between parahippocampal cingulum bundle MR g-ratio and, by inference, conduction velocity, and autobiographical memory recall ability. It may be that vivid, detailed, multimodal, autobiographical memories rely on inter-regional connectivity, particularly that supported by the parahippocampal cingulum bundle, to a greater degree than simpler, more constrained laboratory-based memory tests. This result aligns with previous work involving an fMRI meta-analysis which showed that recall of autobiographical memories and laboratory-based memory stimuli were associated with substantially different neural substrates (McDermott et al., 2009). More generally, our findings add to the increasingly-recognised importance of studying real-world cognition in order to fully characterise brain-behaviour relationships (Maguire, 2001, 2012; Miller et al., 2022; Mobbs et al., 2021; Nastase et al., 2020; Spiers and Maguire, 2007).”

Materials and methods, pp. 28-29:

“Laboratory-based memory tests

Eight laboratory-based memory tasks were also administered to participants. These are standard memory tests that are often used in neuropsychological settings. Tasks were performed and scored in line with their standardised and published protocols.

The ability to recall a short narrative was examined using the immediate and delayed recall tests of the Logical Memory subtest of the Wechsler Memory Scale IV (Wechsler, 2009). Verbal list recall was assessed using the immediate and delayed recall tests of the Rey Auditory Verbal Learning Test (see Strauss et al., 2006). Visuospatial recall was examined using the delayed recall of the Rey–Osterrieth Complex Figure (Rey, 1941). Recognition memory was investigated using the Warrington Recognition Memory Tests for Words and Faces (Warrington, 1984). Finally, participants also underwent the “Dead or Alive” task which probes general knowledge about whether famous individuals have died or are still alive, providing a measure of semantic memory (Kapur et al., 1989).”

Materials and methods, p. 35: “We also investigated the relationship between the MR g-ratio and the magnetisation transfer saturation values and eight laboratory-based memory tests. As with the main analyses, partial correlations were performed between the MR g-ratio and magnetisation transfer saturation values and the outcome measures of the memory tests with age, gender, scanner and the number of voxels in a tract ROI included as covariates, and bootstrapping performed 10,000 times to calculate confidence intervals.”

I am not particularly concerned about correcting for multiple comparisons or arbitrary thresholds for significance. I think that the study is important, novel, and well carried out and that the findings will be important to researchers in several fields. If it is the case that reliable associations between cognitive measures and the MR g-ratio cannot be established in a sample of 217 people then this is really useful information!

Thank you.

I have probably misunderstood the interpretation of the findings with respect to the examples shown in figure 1. It seemed that the authors interpreted their positive correlation between internal details and the g-ratio in the parahippocampal cingulum bundle as evidence for option D – people with better memory recall scores are likely to have larger inner axon diameters and relatively smaller increases in myelin. However, since example 1E involves no change in the g-ratio, I couldn't understand why the finding of no relationship between memory recall ability and g-ratio from the fornix or uncinate fasciculus was not taken as support for option E. It might help if the authors could clarify which situation(s) would lead to them accepting option E as the best explanation of their results.

We are grateful for the opportunity to clarify the situation regarding option E (Figure 2A black arrow, and Figure 2E), and agree it requires further unpacking.

In fact, there are three main scenarios. (1) A decrease in MR g-ratio values. This would suggest that faster conduction velocity is due to greater thickness of the myelin sheath, with the inner axon diameter remaining constant (Figure 2A blue arrow, and Figure 2B). (2) A decrease in MR g-ratio values, but to a lesser extent than that observed in the first scenario. This would suggest that faster conduction velocity is due primarily to greater myelin sheath thickness, but one that is also accompanied by a larger inner axon diameter (Figure 2A red arrow, and Figure 2C). (3) An increase in MR g-ratio values. This would suggest that faster conduction velocity is predominantly due to a larger inner axon diameter, with only small increases in myelin thickness also being present (Figure 2A orange arrow, and Figure 2D).

A fourth scenario also exists in which constant MR g-ratio values could be associated with faster conduction velocity. This would occur when both the inner axon diameter and myelin thickness change proportionally to each other (Figure 2A black arrow, and Figure 2E). However, this scenario could also mean that there is no variation in conduction velocity. One way to increase interpretability in this situation, is by examining scans optimised to assess myelination (such as magnetisation transfer saturation maps). Observing no relationship with the MR g-ratio, but a relationship with magnetisation transfer saturation values, would suggest a proportional change in the underlying myelin and, consequently, variation in conduction velocity. By contrast, no relationship with either the MR g-ratio or magnetisation transfer saturation values would suggest that there was no change in the underlying microstructure, and therefore no variation in conduction velocity. In fact, we observed no associations between the magnetisation transfer saturation values and autobiographical memory recall, suggesting no relationships with myelin or conduction velocity.

We now explicitly consider Figure 2E (black arrow) and its interpretation in the revised manuscript, as detailed below.

Introduction, p. 4: “The conduction velocity of an axon is, therefore, not determined only by the axon diameter, but also by the relationship between the axon diameter and the thickness of the surrounding myelin sheath, a measure known as the g-ratio (Chomiak and Hu, 2009; Rushton, 1951; Schmidt and Knösche, 2019). Specifically, Rushton (1951) derived an equation: ${\displaystyle conductionvelocity\propto d\sqrt{-ln(g)}}$, where d is the inner axon diameter, g is the g-ratio = d/D, and D is the outer fibre (axon plus myelin sheath) diameter. In other words, the g-ratio is computed as the ratio of the inner axon diameter relative to that of the axon plus the myelin sheath that encases it (Figure 1).”

Introduction, p. 5: “Of note, the MR g-ratio has been found to associate well with estimates of axonal conduction velocity (Berman et al., 2019; Drakesmith et al., 2019). Consequently, the MR g-ratio provides a non-invasive MRI method that can associate in vivo structural neuroimaging of humans with axonal conduction velocity.”

Introduction, p. 7: “We note, for completeness, that a fourth scenario also exists where constant MR g-ratio values could be associated with faster conduction velocity. This would occur when both the inner axon diameter and myelin thickness change proportionally to each other (Figure 2A black arrow, and Figure 2E). However, this scenario could also mean that there is no variation in conduction velocity. One way to increase interpretability in this situation, is by examining scans optimised to assess myelination (such as magnetisation transfer saturation values). Observing no relationship with the MR g-ratio, but a relationship with magnetisation transfer saturation values, would suggest a proportional change in the underlying myelin and, consequently, variation in conduction velocity. By contrast, no relationship with either the MR g-ratio or magnetisation transfer saturation values would suggest that there was no change in the underlying microstructure and therefore no variation in conduction velocity.”

Introduction, p. 9: “The relationship of MR g-ratio to the underlying microstructure as outlined in Figure 2, would further guide us as to whether any significant effects were more likely to be associated with the extent of myelination or the size of the inner axon diameter of the fibres in these three white matter tracts. These analyses were augmented by examining whether magnetisation transfer saturation values of the pathways (assessing myelination) were associated with autobiographical memory recall ability.”

Introduction, pp. 10-11: “Focusing on the fornix, uncinate fasciculus and parahippocampal cingulum bundle, we predicted that variations in the MR g-ratio from some or all of these tracts would be associated with autobiographical memory recall ability. Such a finding would, for the first time, suggest a link between variations in white matter tract conduction velocity and individual differences in autobiographical memory recall.

Under the assumption that a significant relationship between the MR g-ratio and autobiographical memory recall represents an association with faster conduction velocity (e.g. Brancucci, 2012; Dicke and Roth, 2016; Miller, 1994; Reed and Jensen, 1992), and aided by the analysis of the magnetisation transfer saturation values, we further sought to evaluate the scenarios presented in Figure 2. Specifically, we asked whether autobiographical memory recall ability was more likely to be associated with inner axon diameters or myelin thickness. A negative relationship between the MR g-ratio and autobiographical memory recall ability (Figure 2A blue and red arrows, and Figures 2B and 2C), along with a positive relationship between magnetisation transfer saturation and autobiographical memory recall ability, would suggest that myelin thickness was more relevant. By contrast, a positive relationship between the MR g-ratio and autobiographical memory recall ability (Figure 2A orange arrow, and Figure 2D) would highlight the potential relevance of inner axon diameters. Observing no associations between the MR g-ratio or magnetisation transfer saturation values and autobiographical memory recall, would suggest no relationships with the underlying microstructure or conduction velocity, and would speak against the possibility of a proportional change in microstructure (Figure 2A black arrow, and Figure 2E).”

Results, p. 16: “To further aid the interpretation of the relationship with MR g-ratio in terms of the underlying axonal microstructure, we also investigated whether the magnetisation transfer saturation values (assessing myelination) of the parahippocampal cingulum bundle were associated with autobiographical memory recall ability. We performed partial correlation analyses with the same covariates as before. Magnetisation transfer saturation values were not significantly related to either internal (r(211) = 0.00, p = 1.0, 95% CI = -0.12, 0.12), or external (r(211) = 0.14, p = 0.048, 95% CI = -0.06, 0.30) details.”

Discussion, p. 20: “The positive relationship we observed between the parahippocampal cingulum bundle MR g-ratio and autobiographical memory recall ability suggests that this effect was associated predominantly with larger inner axon diameter (Figure 2A orange arrow, and Figure 2D). By contrast, had a negative correlation between the MR g-ratio and autobiographical memory recall ability been identified, we could instead have inferred that myelination was the relevant microstructural feature. The lack of relationship between parahippocampal cingulum bundle magnetisation transfer saturation values (optimised to assess myelination) and autobiographical memory recall ability provided further corroboration that myelin may not have been influential in this context.”

Discussion, pp. 22-23: “We also examined two such tracts, the fornix and the uncinate fasciculus, but in both cases no significant relationships between any of our neuroimaging metrics and autobiographical memory recall ability were evident in our large cohort of young healthy adults. As outlined in Figure 2, there was one possible scenario where a null relationship with the MR g-ratio could reflect an association with conduction velocity – if both the inner axon diameter and myelin thickness increased proportionally to each other (Figure 2A black arrow, and Figure 2E). However, we found no associations between autobiographical memory recall and both the MR g-ratio and magnetisation transfer saturation values of the fornix and uncinate fasciculus, which speaks against this explanation.”

The following paper by Hodgetts et al., seems very relevant – the present study fails to replicate their findings related to the fornix.

Hodgetts, C. J., Postans, M., Warne, N., Varnava, A., Lawrence, A. D., and Graham, K. S. (2017). Distinct contributions of the fornix and inferior longitudinal fasciculus to episodic and semantic autobiographical memory. cortex, 94, 1-14.

Thank you for highlighting this paper, which we now reference.

Discussion, p. 23: “In terms of standard DTI parameters, unlike a previous study (Hodgetts et al., 2017), we did not find a relationship between fornix fractional anisotropy and autobiographical memory recall ability, which may be due to the larger sample that we examined.”

Reviewer #2 (Recommendations for the authors):

– Following what I mentioned in the public review, given the lack of any other measure to quantitatively or even qualitatively assess either conduction velocity/delays or axon size, the authors should refrain from directly claiming that larger axons mean better memory recall abilities, which instead is done in the abstract (page 2, "(…) two tract features seemed to favour better memory retrieval – large inner axon diameters (…)" – despite the axon diameters not being measured or estimated), and in the discussion (page 18, "(…) we were able to identify two particular features (…) that might favour better memory recall, namely larger inner axon diameters (…)").

Changes have been made throughout the manuscript to avoid any direct claims of this nature, and to emphasise inferences and associations, where appropriate.

– Related to the previous point, a suggestion that I have is to include, among the exhaustive comparisons performed in the appendix, the ones with the MTsat values. As an absolute measure of myelin, MTsat could further help the authors by providing a complete picture of their results.

As noted in our response to the Reviewer’s Public Review, thank you, we think this is an excellent suggestion. We now include analyses examining relationships between the magnetisation transfer saturation values and autobiographical memory recall ability for all tracts. No significant associations were evident. We also note in the revised manuscript that these findings support the inference that there seems to be no relationship between conduction velocity and autobiographical memory recall for the fornix and uncinate fasciculus. It also aligns with our assumption that here, relationships with the MR g-ratio were indicative of faster, not slower, conduction velocity, and that for the parahippocampal cingulum bundle inner axon diameter might be of more relevance than myelin thickness in relation to autobiographical memory recall ability.

Introduction, p. 7: “We note, for completeness, that a fourth scenario also exists where constant MR g-ratio values could be associated with faster conduction velocity. This would occur when both the inner axon diameter and myelin thickness change proportionally to each other (Figure 2A black arrow, and Figure 2E). However, this scenario could also mean that there is no variation in conduction velocity. One way to increase interpretability in this situation, is by examining scans optimised to assess myelination (such as magnetisation transfer saturation values). Observing no relationship with the MR g-ratio, but a relationship with magnetisation transfer saturation values, would suggest a proportional change in the underlying myelin and, consequently, variation in conduction velocity. By contrast, no relationship with either the MR g-ratio or magnetisation transfer saturation values would suggest that there was no change in the underlying microstructure and therefore no variation in conduction velocity.”

Introduction, p. 9: “The relationship of MR g-ratio to the underlying microstructure as outlined in Figure 2, would further guide us as to whether any significant effects were more likely to be associated with the extent of myelination or the size of the inner axon diameter of the fibres in these three white matter tracts. These analyses were augmented by examining whether magnetisation transfer saturation values of the pathways (assessing myelination) were associated with autobiographical memory recall ability.”

Introduction, p. 11: “Under the assumption that a significant relationship between the MR g-ratio and autobiographical memory recall represents an association with faster conduction velocity (e.g. Brancucci, 2012; Dicke and Roth, 2016; Miller, 1994; Reed and Jensen, 1992), and aided by the analysis of the magnetisation transfer saturation values, we further sought to evaluate the scenarios presented in Figure 2. Specifically, we asked whether autobiographical memory recall ability was more likely to be associated with inner axon diameters or myelin thickness. A negative relationship between the MR g-ratio and autobiographical memory recall ability (Figure 2A blue and red arrows, and Figures 2B and 2C), along with a positive relationship between magnetisation transfer saturation and autobiographical memory recall ability, would suggest that myelin thickness was more relevant. By contrast, a positive relationship between the MR g-ratio and autobiographical memory recall ability (Figure 2A orange arrow, and Figure 2D) would highlight the potential relevance of inner axon diameters. Observing no associations between the MR g-ratio or magnetisation transfer saturation values and autobiographical memory recall, would suggest no relationships with the underlying microstructure or conduction velocity, and would speak against the possibility of a proportional change in microstructure (Figure 2A black arrow, and Figure 2E).”

Results, p. 14: “There were no significant associations between fornix or uncinate fasciculus MR g-ratios or magnetisation transfer saturation values and any of the laboratory-based memory tests (see Appendix 1– tables 5 and 6 for full details, source data are available in Supplementary file 1).”

Results, p. 16: “To further aid the interpretation of the relationship with MR g-ratio in terms of the underlying axonal microstructure, we also investigated whether the magnetisation transfer saturation values (assessing myelination) of the parahippocampal cingulum bundle were associated with autobiographical memory recall ability. We performed partial correlation analyses with the same covariates as before. Magnetisation transfer saturation values were not significantly related to either internal (r(211) = 0.00, p = 1.0, 95% CI = -0.12, 0.12), or external (r(211) = 0.14, p = 0.048, 95% CI = -0.06, 0.30) details.”

Results, p. 18:

“No relationship between microstructure measures of the parahippocampal cingulum bundle and scores on laboratory-based memory tests

Finally, we tested for associations between the key microstructure measures and performance on the eight laboratory-based memory tasks. No relationships with the MR g-ratio or magnetisation transfer saturation values were evident for any task, even when using an uncorrected p < 0.05 threshold (Appendix 1-table 21, source data are available in Supplementary file 2). This suggests that the parahippocampal cingulum bundle MR g-ratio, and by inference conduction velocity, was specifically associated with recall of autobiographical memories from real life rather than performance on the more constrained laboratory-based memory tests.”

Discussion, p. 20: “The positive relationship we observed between the parahippocampal cingulum bundle MR g-ratio and autobiographical memory recall ability suggests that this effect was associated predominantly with larger inner axon diameter (Figure 2A orange arrow, and Figure 2D). By contrast, had a negative correlation between the MR g-ratio and autobiographical memory recall ability been identified, we could instead have inferred that myelination was the relevant microstructural feature. The lack of relationship between parahippocampal cingulum bundle magnetisation transfer saturation values (optimised to assess myelination) and autobiographical memory recall ability provided further corroboration that myelin may not have been influential in this context.”

Discussion, pp. 22-23: “We also examined two such tracts, the fornix and the uncinate fasciculus, but in both cases no significant relationships between any of our neuroimaging metrics and autobiographical memory recall ability were evident in our large cohort of young healthy adults. As outlined in Figure 2, there was one possible scenario where a null relationship with the MR g-ratio could reflect an association with conduction velocity – if both the inner axon diameter and myelin thickness increased proportionally to each other (Figure 2A black arrow, and Figure 2E). However, we found no associations between autobiographical memory recall and both the MR g-ratio and magnetisation transfer saturation values of the fornix and uncinate fasciculus, which speaks against this explanation.”

– It is rather peculiar that the abstract does not mention at all myelin (which in the end is the main protagonist of the study) and that does not briefly define the g-ratio. I understand the authors' focus on the conduction velocity, but these additions would surely benefit a more general audience.

– Related to the previous point, the authors may also consider rethinking the use of "signalling speed", which basically appears only in the title and in the abstract, in favour of "conduction velocity" which is used extensively in the text.

The abstract has now been updated to include reference to myelination and to define the g-ratio. The phrase “signalling speed” has also been replaced in the title with “conduction velocity”. Thank you for these suggestions.

– On page 3, the authors stated: "The scope for signals to travel quickly and efficiently between brain regions could make all the differences for an individual's survival". Although the idea that faster conduction velocity constitutes an evolutionary advantage could be appealing and the peripheral nervous system (PNS) supports the concept extremely well (a very famous example is depicted in Wedel, Acta Palaeontologica 2011), there are several reasons to be cautious with such a statement. In contrast to what happens in the PNS (where faster propagation reflects in faster sensory processing and faster motor responses), in the central nervous system conduction phenomena serve the purpose of guaranteeing synchrony in the activity of different neuronal populations (Seidtl, Neuroscience 2014; Gollo et al., NeuroImage 2017; Pariz et al., Plos Computational Biology 2021) and the role of delays could be more subtle. Unless the authors can integrate the statement with some references to experimental evidence, I would remove such a statement.

We agree, and have now removed this statement from the Introduction.

– Still on page 3, the authors mentioned that the conduction velocity "is dependent upon two biological features – the axonal diameter and the presence and thickness of a myelin sheath". This picture is too approximative as experimental work (Arancibia-Carcamo et al., eLife 2017) and simulations (Drakesmith et al., NeuroImage 2019) have shown how several parameters influence conduction velocity, potentially even with a larger impact than the g-ratio. It would be appropriate for a more general audience to add more details. Similarly, on page 4 it would help to refer to saltatory propagation when explaining the effect of myelin as an insulator.

In the revised manuscript we now include additional details about how the conduction velocity of an axon is dependent upon its physiology, and what factors are measurable in vivo in humans. We also refer to saltatory propagation when explaining the effect of myelin as an insulator. We agree that these additions provide a more informative context.

Introduction, pp. 3-4:

“The conduction velocity of an axon is dependent upon the axon diameter, the presence and thickness of a myelin sheath, the distance between the nodes of Ranvier (periodic gaps in the myelin that facilitate action potential propagation), inter-nodal spacing, and electrical properties of the axonal and myelin membranes (Arancibia-Cárcamo et al., 2017; Drakesmith et al., 2019; Gasser and Grundfest, 1939; Hursh, 1939; Huxley and Stämpeli, 1949; Rushton, 1951). A number of these features are not yet measurable in humans in vivo. However, seminal electrophysiological work has derived a relationship between axon morphology and conduction velocity using only axon diameter and myelin sheath thickness (Rushton, 1951). These two metrics are particularly key because a larger axon diameter results in less resistance to the action potential ion flow, resulting in faster conduction velocity. The presence of a myelin sheath around an axon is beneficial in two ways. First, the myelin sheath acts like an electrical insulating layer, reducing ion loss and preserving the action potential. Second, the presence of unmyelinated gaps in the myelin sheath (the nodes of Ranvier) enables a process called saltatory propagation to take place. As the majority of the axon is wrapped in myelin, the nodes of Ranvier are the only locations where action potentials can occur. This increases the strength of electrical signals because all the ions gather at these nodes instead of being dispersed along the length of the axon. Stronger action potentials are therefore sent along the myelinated portion of the axon at higher speeds, with this signal being boosted on arrival at the next node of Ranvier by another action potential, which helps to maintain a fast conduction velocity.

The conduction velocity of an axon is, therefore, not determined only by the axon diameter, but also by the relationship between the axon diameter and the thickness of the surrounding myelin sheath, a measure known as the g-ratio (Chomiak and Hu, 2009; Rushton, 1951; Schmidt and Knösche, 2019). Specifically, Rushton (1951) derived an equation: ${\displaystyle conductionvelocity\propto d\sqrt{-ln(g)}}$, where d is the inner axon diameter, g is the g-ratio = d/D, and D is the outer fibre (axon plus myelin sheath) diameter. In other words, the g-ratio is computed as the ratio of the inner axon diameter relative to that of the axon plus the myelin sheath that encases it (Figure 1).”

– On page 7, there is a bit of a leap between the section on the uses and potentials of the MRI-based g-ratio and the proposed application in autobiographical memory. It would help to provide a couple of sentences to link those together or to rearrange a bit the paragraph in-between pages 7 and 8, where at the very end the case for g-ratio and memory is made.

We have now finessed the text further to provide a smoother link.

Introduction, p. 7: “The MR g-ratio has, therefore, the potential to provide a number of novel insights into human cognition. One area where the MR g-ratio may be particularly helpful is in probing individual differences. Our particular interest is in the ability to recall past experiences from real life, known as autobiographical memories.”

– The considerations in figure 1 made on pages 4-5 are at the moment very hard without referring to an actual relationship between conduction velocity, axonal diameter, and g-ratio. Even though the authors are not actually computing conduction velocity, it would help to refer to established formulas such as the one from Rushton (Rushton, Journal of Physiology 1951) or the one form Waxman and Bennet (Waxman and Bennet, Nature New Biology 1972).

We now refer explicitly to the Rushton (1951) formulae for calculating conduction velocity.

Introduction, p. 4: “The conduction velocity of an axon is, therefore, not determined only by the axon diameter, but also by the relationship between the axon diameter and the thickness of the surrounding myelin sheath, a measure known as the g-ratio (Chomiak and Hu, 2009; Rushton, 1951; Schmidt and Knösche, 2019). Specifically, Rushton (1951) derived an equation: ${\displaystyle conductionvelocity\propto d\sqrt{-ln(g)}}$, where d is the inner axon diameter, g is the g-ratio = d/D, and D is the outer fibre (axon plus myelin sheath) diameter. In other words, the g-ratio is computed as the ratio of the inner axon diameter relative to that of the axon plus the myelin sheath that encases it (Figure 1).”

– I noticed that in Figures 4 and 5 the scatterplots are based on the residuals rather than the actual g-ratio and neurite dispersion values. Can I ask what is the reason behind this choice? This representation makes actually harder to get a visual sense of the dynamics of the reported relationships.

The scatterplots are based on the residuals because we included a number of covariates – age, gender, scanner, number of voxels in the ROI – in the correlation analyses. Plotting the actual g-ratio (or neurite dispersion) values against autobiographical memory recall would not present an accurate plot of the relationship, as calculated with the conservative statistics we employed.

– On page 18, the authors mentioned how these results can provide a new perspective on the underlying neural substrate of autographical memory, and stated: "This is especially welcome given the dearth of consistent findings linking hippocampal grey matter volume or microstructure with autobiographical memory recall ability in the healthy population". Can the authors expand their argument here? I am missing the link between the current findings (on white matter microstructure) with previous ones on grey matter features.

This sentence was simply meant to highlight the fact that we currently lack a clear biological explanation for individual differences in autobiographical memory recall in the healthy population. We have now re-worded this to better reflect the point being made.

p. 19: “These results offer a new perspective on individual differences in autobiographical memory recall ability. This is especially welcome given the current lack of a clear biological explanation for such variations in the healthy population (Clark et al., 2020; Clark et al., 2021a; LePort et al., 2012; Maguire et al., 2003; Palombo et al., 2018; Van Petten, 2004).”

– On page 22, there is a reference to the CONNECTOM MRI scanners (misspelled as "connectome MRI scanners") – as the general audience is not familiar with those it would help to add a brief sentence explaining what they are and maybe a reference (for example, Jones et al., NeuroImage 2018).

We have now included an explanation of Connectom MRI scanners.

Introduction, pp. 5-6: “While measures of myelination can be obtained using conventional MRI scanners (e.g. via magnetisation transfer saturation maps), estimates of inner axon diameter cannot, and instead require ultra-strong gradient systems – 300 mT/m compared to the 10s of mT/m of conventional MRI scanners (Jones et al., 2018; Veraart et al., 2020). However, only four ultra-strong gradient Connectom MRI scanners exist in the world. Consequently, measuring the MR g-ratio with conventional MRI scanners can be useful in guiding inferences about inner axon diameter.”

– Can the authors comment on the advantages of the proposed ROI-based approach compared with a tract-based approach?

We now mention the advantages of the ROI compared to a tract-based approach in the Materials and methods.

Materials and methods, p. 34: “We used the well-established ROI approach (e.g. Ellerbrock and Mohammadi, 2018; Memel et al., 2020) rather than, for example, a tract-based pipeline, in order to reduce the influence of seed and target region selection. This is because small changes to these selections can result in different tracts being identified. We also wanted to avoid the inclusion of excess grey matter in the tracts themselves, because it is not possible to estimate the MR g-ratio in grey matter tissue. In addition, an ROI approach reduced the potential for false positives in comparison to performing tract-based voxel-wise analyses.”

– I may have missed it, but I am not sure of what is the reason for the fornix tract being "not probabilistic" (page 29) – are not all the tract ROIs obtained from the John Hopkins probabilistic atlas?

Thank you for noting this. The fornix was defined using the ICBM-DTI-81 white-matter labels atlas (Mori et al., 2008, NeuroImage), which is a non-probabilistic atlas. The uncinate fasciculus and parahippocampal cingulum bundle were defined using the Johns Hopkins probabilistic atlas. This latter atlas does not include the fornix. The manuscript text has been updated accordingly.

Materials and methods, p. 33: “Bilateral tract ROIs were defined using the Johns Hopkins probabilistic white matter tractography atlas (Hua et al., 2008), with the exception of the fornix which was defined using the ICBM-DTI-81 white-matter labels atlas (Mori et al., 2008) as the fornix is not available in the probabilistic atlas.”

Figure 3 legend: “The three white matter tracts of interest, given their relationship with the hippocampal region. The fornix was defined using the ICBM-DTI-81 white-matter labels atlas (Mori et al., 2008). The uncinate fasciculus and parahippocampal cingulum bundle were defined using the Johns Hopkins probabilistic white matter tractography atlas (Hua et al., 2008), with the minimum probability threshold set to 25%”.

– I could not find any mention of the main methodological limitations (e.g., the use of signal fractions rather than volume fractions, the bias towards larger axons, and the need for single-point calibration). There is only a brief mention in the appendix on page 43. I think these limitations should be part of the discussion to put these results in the right context.

The main methodological limitations are now outlined in the revised manuscript.

Discussion, p. 25: “Nevertheless, as with all neuroimaging techniques, methodological limitations need to be considered when measuring the MR g-ratio (see Campbell et al., 2018; Mohammadi and Callaghan, 2021 for in depth methodological reviews). First, the MR g-ratio is an area-weighted average of all microscopic axons in an MRI-voxel that is slightly weighted towards larger axons. Second, MR proxies are required to estimate myelin and axonal volumes for the calculation of the MR g-ratio. Multiple methodologies are available to estimate myelin and axonal volumes, with no consensus yet reached as to the best combination (Mohammadi and Callaghan, 2021). Third, a calibration step is required to more closely align the MR proxies with the estimated volume fractions. Here, we used the standard single-point calibration method to estimate the slope of the myelin-based proxy, assuming that the offset can be neglected. Future work may be able to improve these calibrations. But, of note, since a non-negligible offset in the myelin-based proxy can increase the error in the MR g-ratio, any calibration improvements would likely serve to increase the observed correlation. Fourth, although the model of the MR g-ratio allows for fibre dispersion (Stikov et al., 2015), this is not accounted for in mapping to conduction velocity. Going forward, more advanced models may help to further elaborate on these relationships. Finally, our myelin measure can be influenced by any factor, cognitive or demographic, that leads to a difference in absolute myelination. In principle, unexplained variance could obscure a true underlying relationship with autobiographical memory. To mitigate this, our analyses controlled for age, gender and any potential scanner-related differences. The MR g-ratio, however, is less affected by this limitation because it is a relative measure and does not depend on absolute myelin content but on the balance between absolute myelin and axonal volumes.”

Reviewer #3 (Recommendations for the authors):

I have some suggestions the authors may consider, presented in chronological order.

Title

The title suggests a causal relationship, while this is based on an association. Maybe better to use a neutral word, i.e. replace "influences" with e.g. "is associated with" or something similar?

As I understand MR g-ratio, it reflects structural aspects of nerve fibers, and although probably related to conduction velocity/ signaling speed, it is not a direct measure of this. Hence, I believe the title is overstating this relationship by referring to "signalling speed" directly. I fully understand the need for short and focused titles, but I believe this is a bit misleading. The same point applies to a couple of other places in the manuscript too. Of course fine to discuss this as an interpretation.

The title has now been revised to replace “signalling speed” with Reviewer #2’s suggestion of “conduction velocity”, and Reviewer #3’s suggestion to replace “influences” with “associated with”.

Title, p. 1: “Conduction velocity along a key white matter tract is associated with autobiographical memory recall ability”

Any references to causal relationships have also been removed throughout the manuscript.

Abstract

Not all readers are probably familiar with the MR g-ratio, so maybe it can be defined in the abstract?

A definition of the g-ratio has now been included.

Abstract, p. 2: “Inferences about conduction velocity can now be made in vivo in humans using a measure called the magnetic resonance (MR) g-ratio. This is the ratio of the inner axon diameter relative to that of the axon plus the myelin sheath that encases it.”

Since this is a cross-sectional study, I suggest avoiding phrases that imply change, e.g. "increased MR g-ratio". Maybe replace it with something neutral, such as "higher MR g-ratio"? The same is the case in several places in the manuscript, e.g. ln 195.

As with the title, the results reflect associations, not causation. This does not render them less interesting, but I would recommend replacing "favour better memory …" with e.g. "is associated with better memory …". I also wonder whether "offer a new perspective on drivers of individual differences in memory recall ability" is overstating a bit, since (a) we don't know whether g-ratio (or signal speed) is driving anything, and (b) this is a rather well-established theory (as also explained in the introduction, and it is stated ln 44 "it is perhaps no surprise that speed may be of the essence")?

Throughout the manuscript we have replaced phrases that imply change with more neutral associative statements, and generally tempered the language.

Introduction

ln 51-54: Why was aging discussed? Reading the introductory paragraph, I expected aging to be part of the study, which it was not.

The literature examining conduction velocity in relation to individual differences is relatively sparse, especially as most work in this area is performed in non-human animals, and involves ageing, which is why we cited it. We have now removed the aging-related summary sentence at the end of that paragraph to prevent confusion.

Ln 123-125: "While some healthy individuals can recollect decades-old autobiographical memories with great richness and clarity, others struggle to recall what they did last weekend" Yes, but as the authors surely know, this reflects different memory systems or processes, and therefore can happen within the same individual too, not only as a feature of inter-individual differences.

We do not deny that variability in recall occurs within the same person as well across people. However, the current study was set up to examine individual differences across people, and the two references cited in relation to that statement (LePort et al., 2012; Palombo et al., 2015) also refer to inter-individual differences.

Ln 195-200: Figure 1 clearly illustrates different scenarios, which is very useful. However, the hypothesis that better autobiographical memory is related to more myelin could also be consistent with no relationship with the MR g-ratio if the axon diameter is larger in participants with a good memory, right? And vice versa for the hypothesis about axon diameter? I don't see how the ratio measure can distinguish between the different accounts since the measures per definition are related to two possibly independent properties which both can be of importance? The authors multiple times mention that other relevant metrics, such as those derived from DTI, lack biological specificity. But the MR g-ratio measure seems also to be unspecific in this regard.

As we now note in the revised manuscript, the conduction velocity of an axon is not determined by the axon diameter or myelin sheath thickness individually, but by the relationship between them – the g-ratio. Given that the MR g-ratio encompasses information about the inner axon diameter and myelin thickness, the identification of statistically significant relationships with the MR g-ratio can help to guide as to which of these two features might be associated variations in conduction velocity. One way to further increase interpretability is by examining scans optimised to assess myelination (such as magnetisation transfer saturation values), as suggested by Reviewer #2. We observed no relationship between autobiographical memory recall ability and magnetisation transfer saturation values. This finding is supportive of the parahippocampal cingulum bundle inner axon diameter being of potentially more relevance than myelin thickness. Consequently, by measuring the MR g-ratio and magnetisation transfer saturation, we can make biologically-relevant inferences about the relationship between autobiographical memory recall ability and conduction velocity. This is in contrast to metrics derived from DTI. In the revised manuscript, we now explain that fractional anisotropy, for example, is a very general measure. Variations in fractional anisotropy values can occur for numerous reasons, including, but not limited to, changes in the extent of myelination, axon coherence, axon density, the level of astrocytes, and so forth.

Introduction, p. 6: “Adjudicating between the possible influence of inner axon diameter or myelination relies on knowledge, or an assumption, about whether the associated change in conduction velocity is faster or slower. This is because significant associations with the MR g-ratio only indicate the existence of a relationship with conduction velocity but not the direction. As noted previously, faster conduction velocity is often held to promote better cognition (e.g. Brancucci, 2012; Dicke and Roth, 2016; Miller, 1994; Reed and Jensen, 1992). Therefore, in Figure 2A, inner axon diameter, myelin thickness and (MR) g-ratio are plotted together to illustrate how different changes in the MR g-ratio are related to the underlying microstructural properties, given a faster conduction velocity. Myelin thickness is represented by the gradient in background colour and contours, with thinnest myelin at the bottom right, and thickest on the top left. The direction of the arrows describes the change in g-ratio for the microstructural variations presented. There are three main scenarios. (1) A decrease in MR g-ratio values. This would suggest that faster conduction velocity is due to greater thickness of the myelin sheath, with the inner axon diameter remaining constant (Figure 2A blue arrow, and Figure 2B). (2) A decrease in MR g-ratio values, but to a lesser extent than that observed in the first scenario. This would suggest that faster conduction velocity is due primarily to greater myelin sheath thickness, but one that is also accompanied by a larger inner axon diameter (Figure 2A red arrow, and Figure 2C). (3) An increase in MR g-ratio values. This would suggest that faster conduction velocity is predominantly due to a larger inner axon diameter, with only small differences, if any, in myelin thickness being present (Figure 2A orange arrow, and Figure 2D).”

Introduction, pp. 9-10: “For completeness, the commonly reported physical parameters from standard diffusion tensor imaging (DTI; e.g. fractional anisotropy and mean diffusivity; Basser, 1995) that are often derived from diffusion data were also computed (Oeschger et al., 2021). However, these metrics lack biological specificity (Jensen and Helpern, 2010; Jones et al., 2013a) and, consequently, could not speak to our research questions. For example, fractional anisotropy is a very general measure. Variations in fractional anisotropy values can occur for numerous reasons, including, but not limited to, changes in the extent of myelination, axon coherence, axon density, and the level of astrocytes.”

Discussion, pp. 19-20: “A key property of the MR g-ratio is that it provides a non-invasive methodology with which to relate in vivo structural neuroimaging of humans with axonal conduction velocity (Berman et al., 2019; Drakesmith et al., 2019). Moreover, the identification of a statistically significant relationship with the MR g-ratio can help to guide as to which of inner axon diameter or myelin thickness is more likely to be associated with variations in conduction velocity (Caeyenberghs et al., 2016; Kaller et al., 2017; Lakhani et al., 2016; Waxman, 1980; Xin and Chan, 2020).

Our inferences rested upon the assumption that a significant relationship with the MR g-ratio reflected faster conduction velocity, given previous work suggesting that faster conduction velocity might be associated with better cognition (e.g. Brancucci, 2012; Dicke and Roth, 2016; Miller, 1994; Reed and Jensen, 1992). As shown in Figure 2, there are three main scenarios describing how changes in the MR g-ratio are related to the underlying microstructural properties, given a faster conduction velocity. The positive relationship we observed between the parahippocampal cingulum bundle MR g-ratio and autobiographical memory recall ability suggests that this effect was associated predominantly with larger inner axon diameter (Figure 2A orange arrow, and Figure 2D). By contrast, had a negative correlation between the MR g-ratio and autobiographical memory recall ability been identified, we could instead have inferred that myelination was the relevant microstructural feature. The lack of relationship between parahippocampal cingulum bundle magnetisation transfer saturation values (optimised to assess myelination) and autobiographical memory recall ability provided further corroboration that myelin may not have been influential in this context. Greater myelination is often held to be a prominent influence on behavioural and cognitive performance (Caeyenberghs et al., 2016; Fields and Bukalo, 2020; Kaller et al., 2017; Lakhani et al., 2016; Waxman, 1980; Xin and Chan, 2020). By contrast, our results highlight the potentially important role that the inner axon diameter could be playing in autobiographical memory recall.”

If axon diameter and myelin thickness can be estimated separately, why not use them directly in the analyses instead of the MR g-ratio? There may be a methodological reason for this, but I think this could be more clearly explained as MR g-ratio is not often reported in the literature and so will be novel to many readers.

As we now note in the revised manuscript, the conduction velocity of an axon is not determined by the axon diameter or myelin sheath thickness individually, but by the relationship between them – the g-ratio. The MR g-ratio, therefore, is the only measure able to provide a link with conduction velocity.

While measures of myelination can be obtained using conventional MRI scanners (e.g. via the magnetisation transfer saturation values that we also acquired), estimates of inner axon diameter cannot, and instead require ultra-strong gradient systems – 300 mT/m in comparison to the 10s of mT/m of conventional MRI scanners (Jones et al., 2018; Veraart et al., 2020). However, only four ultra-strong gradient Connectom MRI scanners exist in the world. Consequently, measuring the MR g-ratio and interpreting these results together with magnetisation transfer saturation values can be useful in guiding inferences about inner axon diameter with conventional MRI scanners.

Introduction, p. 4: “The conduction velocity of an axon is, therefore, not determined only by the axon diameter, but also by the relationship between the axon diameter and the thickness of the surrounding myelin sheath, a measure known as the g-ratio (Chomiak and Hu, 2009; Rushton, 1951; Schmidt and Knösche, 2019). Specifically, Rushton (1951) derived an equation: ${\displaystyle conductionvelocity\propto d\sqrt{-ln(g)}}$, where d is the inner axon diameter, g is the g-ratio = d/D, and D is the outer fibre (axon plus myelin sheath) diameter. In other words, the g-ratio is computed as the ratio of the inner axon diameter relative to that of the axon plus the myelin sheath that encases it (Figure 1).”

Introduction, p. 5: “Of note, the MR g-ratio has been found to associate well with estimates of axonal conduction velocity (Berman et al., 2019; Drakesmith et al., 2019). Consequently, the MR g-ratio provides a non-invasive MRI method that can associate in vivo structural neuroimaging of humans with axonal conduction velocity.”

Introduction, p. 5-6: “Given that the MR g-ratio encompasses information about the inner axon diameter and myelin thickness, the identification of statistically significant relationships with the MR g-ratio can also guide as to which of these two features might be influencing variations in conduction velocity (Caeyenberghs et al., 2016; Kaller et al., 2017; Lakhani et al., 2016; Waxman, 1980; Xin and Chan, 2020). This is particularly useful in relation to the inner axon diameter. While measures of myelination can be obtained using conventional MRI scanners (e.g. via magnetisation transfer saturation maps), estimates of inner axon diameter cannot, and instead require ultra-strong gradient systems – 300 mT/m compared to the 10s of mT/m of conventional MRI scanners (Jones et al., 2018; Veraart et al., 2020). However, only four ultra-strong gradient Connectom MRI scanners exist in the world. Consequently, measuring the MR g-ratio with conventional MRI scanners can be useful in guiding inferences about inner axon diameter.”

Discussion, pp. 19-20: “A key property of the MR g-ratio is that it provides a non-invasive methodology with which to relate in vivo structural neuroimaging of humans with axonal conduction velocity (Berman et al., 2019; Drakesmith et al., 2019). Moreover, the identification of a statistically significant relationship with the MR g-ratio can help to guide as to which of inner axon diameter or myelin thickness is more likely to be associated with variations in conduction velocity (Caeyenberghs et al., 2016; Kaller et al., 2017; Lakhani et al., 2016; Waxman, 1980; Xin and Chan, 2020).”

Relatedly, although there are methods papers on this topic, maybe a brief justification of how can ratios between these small compartments on the micrometer scale can be calculated from voxels on the millimeter scale already in the introduction?

We now briefly expand this explanation in the revised manuscript.

Introduction, p. 5: “Until recently, g-ratio measurements were restricted to invasive studies in non-human animals. However, by combining diffusion magnetic resonance imaging (MRI) with quantitative structural MRI scans optimised to assess myelination (e.g. magnetisation transfer saturation; Weiskopf et al., 2013), it is now possible to estimate the g-ratio in vivo in humans across the whole brain (Drakesmith et al., 2019; Mohammadi et al., 2015; Mohammadi and Callaghan, 2021; Stikov et al., 2015). This is achieved by measuring an aggregate g-ratio, which is an area-weighted ensemble average across a voxel of an underlying distribution of microscopic g-ratios of axons (Stikov et al., 2015; West et al., 2016).”

Results

I understand that the authors don't want to provide the full description of results for FA and MD in the main text, but it would be convenient to report a brief summary of the main results also in the text. As far as I can see from Appendix 1, none even approach significance for fornix and uncinated fasciculus, so maybe at least state that?

Brief summaries of the standard DTI parameters are now provided in the Results section, and are mentioned in the Discussion.

Results, pp. 13-14: “Standard DTI parameters (e.g. fractional anisotropy and mean diffusivity) were also extracted from the fornix and uncinate fasciculus. None of the standard DTI parameters, from either tract, were significantly associated with autobiographical memory recall ability, even when using an uncorrected p < 0.05 threshold (see Appendix 1–tables 1-4 for full details, source data are available in Supplementary file 1).”

Results, pp. 17-18:

“Standard DTI parameters of the parahippocampal cingulum bundle and autobiographical memory recall ability

Partial correlations using the same covariates as the previous analyses were also performed using the standard DTI parameters. As these metrics lack biological specificity (Jensen and Helpern, 2010; Jones et al., 2013a), and could not speak to our research questions, the results are summarised here, with full details in Appendix 1–tables 19 and 20 and Appendix 1–figures 9 and 10, and the source data are available in Supplementary file 2.

Two significant relationships were identified between the standard DTI parameters and autobiographical memory recall ability. First, there was a positive correlation between fractional anisotropy and the number of internal details (Appendix 1–figure 9A; r(211) = 0.20, p = 0.003, 95% CI = 0.07, 0.32). Second, a positive correlation was evident between diffusivities parallel and the number of internal details (Appendix 1–figure 10A; r(211) = 0.19, p = 0.005, 95% CI = 0.06, 0.32). As with the microstructure metrics, these relationships were specific to internal details with no relationships identified with external details for either fractional anisotropy (r(211) = -0.06, p = 0.39, 95% CI = -0.19, 0.07) or diffusivities parallel (r(211) = -0.048, p = 0.49, 95% CI = -0.19, 0.10). No significant relationships were found between mean diffusivity, mean kurtosis or diffusivities perpendicular and either internal or external details.”

Discussion, p. 22: “There have only been a small number of previous studies investigating white matter tracts and individual differences in autobiographical memory recall ability in healthy people. These focused on standard DTI parameters, such as fractional anisotropy and mean diffusivity, and not the MR g-ratio. Two such studies were suggestive of a relationship between parahippocampal cingulum bundle fractional anisotropy and autobiographical memory recall (Irish et al., 2014; Memel et al., 2020). Our analyses of the standard DTI parameters, therefore, replicated these previous results.”

Discussion, p. 23: “In terms of standard DTI parameters, unlike a previous study (Hodgetts et al., 2017), we did not find a relationship between fornix fractional anisotropy and autobiographical memory recall ability, which may be due to the larger sample that we examined.”

This may be a matter of taste, but I would suggest moving the section on why the parahippocampal cingulum bundle should be important for autobiographical memory from the Results section (ln 236 and onwards) to the introduction or the discussion.

We understand the Reviewer’s point, and did consider this idea before the original submission. We have elected to keep this section within the Results to act as a primer for readers when they reach the parahippocampal cingulum bundle findings, to remind them about why this tract may be especially relevant. However, consideration of the parahippocampal cingulum bundle and its possible role in autobiographical memory recall is also included in both the Introduction and Discussion.

Ln 279: A positive relationship between MR g-ratio and memory is interpreted as reflecting larger inner axon diameters. Ref Figure 1, this can also be due to thinner myelin sheets, right, as it is a ratio measure?

A positive relationship between MR g-ratio and autobiographical memory recall would only reflect thinner myelin sheaths if the conduction velocity was slower (rather than faster, which is the basis of our interpretations). In addition, we have now also included analyses with a measure of myelination (magnetisation transfer saturation) and found no relationship between myelin and autobiographical memory recall ability. This supports our suggestion that the MR g-ratio finding might be related to inner axon diameter rather than myelination, and consequently also the assumption of faster rather than slower conduction velocity.

Multiple comparison correction was mentioned for some analyses above, was this done here two, and if so, how? Would be useful to state explicitly.

Multiple comparison correction was performed for all partial correlation analyses using the Bonferroni method; dividing α = 0.05 by 3 (because there were three tracts of interest). This means that associations with a two-sided p-value < 0.017 were considered significant. This information is included in the Materials and methods on p. 35.

Inspecting the Appendix, it seems that the relationships between autobiographic memory and PCB FA and MD were also significant, with r-values that were at least as large as for MR g-ratio. Why not include this in the main text? These are metrics of interest to many researchers.

Brief summaries of the standard DTI parameters are now provided in the Results section, and are mentioned in the Discussion.

Results, pp. 17-18:

“Standard DTI parameters of the parahippocampal cingulum bundle and autobiographical memory recall ability

Partial correlations using the same covariates as the previous analyses were also performed using the standard DTI parameters. As these metrics lack biological specificity (Jensen and Helpern, 2010; Jones et al., 2013a), and could not speak to our research questions, the results are summarised here, with full details in Appendix 1–tables 19 and 20 and Appendix 1–figures 9 and 10, and the source data are available in Supplementary file 2.

Two significant relationships were identified between the standard DTI parameters and autobiographical memory recall ability. First, there was a positive correlation between fractional anisotropy and the number of internal details (Appendix 1–figure 9A; r(211) = 0.20, p = 0.003, 95% CI = 0.07, 0.32). Second, a positive correlation was evident between diffusivities parallel and the number of internal details (Appendix 1–figure 10A; r(211) = 0.19, p = 0.005, 95% CI = 0.06, 0.32). As with the microstructure metrics, these relationships were specific to internal details with no relationships identified with external details for either fractional anisotropy (r(211) = -0.06, p = 0.39, 95% CI = -0.19, 0.07) or diffusivities parallel (r(211) = -0.048, p = 0.49, 95% CI = -0.19, 0.10). No significant relationships were found between mean diffusivity, mean kurtosis or diffusivities perpendicular and either internal or external details.”

Discussion, p. 22: “There have only been a small number of previous studies investigating white matter tracts and individual differences in autobiographical memory recall ability in healthy people. These focused on standard DTI parameters, such as fractional anisotropy and mean diffusivity, and not the MR g-ratio. Two such studies were suggestive of a relationship between parahippocampal cingulum bundle fractional anisotropy and autobiographical memory recall (Irish et al., 2014; Memel et al., 2020). Our analyses of the standard DTI parameters, therefore, replicated these previous results.”

You could also run a multiple regression analysis, and test whether each metric gives unique information or whether they are redundant. I think this would be a very useful addition to the manuscript, regardless of the outcome. If MR g-ratio is still significantly related to autobiographical memory, this would strengthen the main argument in the paper. If not, we will know more about the interrelationships between these different tract metrics, which also will be very useful.

Unfortunately, it is not possible to implement this suggestion. Each of the metrics available has varying specificity in relation to the underlying microstructure. Fractional anisotropy, for example, is a very general measure that simply describes the directionality of water molecule movement. High fractional anisotropy values can occur, therefore, for a variety of reasons including increased myelination, axon coherence and axon density. This means that fractional anisotropy will always be able to explain more of the variance than the more specific measures. However, this does not mean it has greater explanatory power, because fractional anisotropy cannot inform about the actual variations in microstructure, while the more specific metrics can.

To illustrate, in the current study, we see relationships between autobiographical memory recall and fractional anisotropy, axonal coherence and the MR g-ratio, and no relationship with axonal density. As the fractional anisotropy metric is the most general measure, placing the four measures together into a regression model results in fractional anisotropy explaining the majority of the model variance. However, even if we had observed instead, for example, relationships between autobiographical memory recall and fractional anisotropy, axonal density and the MR g-ratio, and no relationship with axonal coherence, exactly the same result would emerge from the regression analysis. Fractional anisotropy would again seem to explain the majority of the model variance, even though different inferences would be drawn from investigating the individual measures alone.

Discussion

Ln 336: Is it correct to say that the MR g-ratio – memory correlation represents "neural instantiation of autobiographical memory"? To me, it seems to overstate the results, as this is an association, and we do not know whether there is any causality involved. The same goes for the last part of the sentence "drivers of individual differences in recall ability". I think the results are of sufficient interest, without such statements.

These statements have now been removed.

Methods, Participants

Ln 456-457: "People with hobbies or vocations known to be associated with the hippocampus (e.g. licenced London taxi drivers) were excluded." Could you please be a little more specific, most activities will be associated with the hippocampus? How many were excluded based on this?

Additional information has now been included in the revised manuscript.

Materials and methods, pp. 26-27: “Our aim was to assess people from the general population who would not be classed as having extreme expertise on classic hippocampal tasks, as this could affect hippocampal structure (Maguire et al., 2000; Woollett and Maguire, 2011). Consequently, people with vocations such as taxi driving (or those training to be taxi drivers), ship navigators, aeroplane pilots, or those with regular hobbies including orienteering, or taking part in memory sports and competitions, were excluded. Of the approximately 2000 people who contacted us, 23 were explicitly excluded on this basis.”

[Editors’ note: what follows is the authors’ response to the second round of review.]

The authors put an amazing effort into taking into account the reviewers' comments and into improving the paper with further details and analysis. The finding that magnetization transfer saturation does not correlate with the score of interest while the g-ratio instead does is quite interesting -

Thank you.

I think that for the sake of fully characterizing the relationship what is missing at the moment is correlating the isotropic volume fraction with the recall score: since the g-ratio is computed using MTsat, Viso and Vic, double-checking that a correlation does not exist with Viso is the last step needed to show that indeed is the combination of all these measures that is meaningful rather than the single measures themselves.

Author response: Thank you for this suggestion. The correlation between autobiographical memory recall ability and Viso (isotropic volume fraction) has now been computed and included in the manuscript (and the associated data are now included in the relevant supplementary file). As expected, no significant relationship was observed, confirming that it is the combination of the measures, in the form of the MR g-ratio, that is meaningful rather than any single measure alone. We have modified the text as follows:

Results, pp. 17-18: “For completeness, we also examined a third measure from the NODDI biophysical model, the isotropic volume fraction. This models the space occupied by cerebrospinal fluid in a voxel, and therefore no relationship with autobiographical memory recall ability was expected, and none was found. Partial correlations showed that the isotropic volume fraction of the parahippocampal cingulum bundle (mean = 0.04, SD = 0.03) was not significantly related to either internal (r(211) = -0.03, p = 0.71, 95% CI = -0.14, 0.10), or external (r(211) = 0.02, p = 0.77, 95% CI = -0.11, 0.15) details. This confirms that it is the combination of these measures, in the form of the MR g-ratio, that is meaningful, rather than any single measure alone.”

Discussion, p. 25: “Relationships with autobiographical memory recall ability were also specific to the MR g-ratio, and were not evident for any of its components (magnetisation transfer saturation, neurite density or isotropic volume fraction), suggesting that it is the combination of these measures, in the form of the MR g-ratio, that is meaningful rather than any single measure alone.”

Author details

1. Ian A Clark

   Wellcome Centre for Human Neuroimaging, Department of Imaging Neuroscience, UCL Queen Square Institute of Neurology, University College London, London, United Kingdom

   Contribution

   Conceptualization, Formal analysis, Investigation, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-5678-2190

2. Siawoosh Mohammadi

   Institute of Systems Neuroscience, University Medical Centre Hamburg-Eppendorf, Hamburg, Germany

   Contribution

   Formal analysis, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-1311-9636

3. Martina F Callaghan

   Wellcome Centre for Human Neuroimaging, Department of Imaging Neuroscience, UCL Queen Square Institute of Neurology, University College London, London, United Kingdom

   Contribution

   Formal analysis, Writing - review and editing

   Competing interests

   No competing interests declared

4. Eleanor A Maguire

   Wellcome Centre for Human Neuroimaging, Department of Imaging Neuroscience, UCL Queen Square Institute of Neurology, University College London, London, United Kingdom

   Contribution

   Conceptualization, Formal analysis, Supervision, Funding acquisition, Methodology, Writing - original draft, Writing - review and editing

   For correspondence

   e.maguire@ucl.ac.uk

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-9470-6324

• 668

  Page views

• 138

  Downloads

• 1

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.