Much like digital cameras record images using a grid of tiny pixels, our own visual experience results from the activity of many neurons, each with its own receptive field. A neuron’s receptive field is the area of visual space – the scene in front of our eyes – to which that neuron responds. But whereas digital pixels have fixed locations, the receptive fields of neurons do not. If we switch our attention to a different area of the scene in front of us, visual neurons move their receptive fields to cover that area instead. We do not need to move our eyes for this to happen, just the focus of our attention.

Moving receptive fields in this way enables the visual system to generate more detailed vision at the new attended location. Unlike a digital camera, the brain is thus much more than a passive recording device. But does the movement of receptive fields also depend on what we are attending to at a given location? Paying attention to tiny details, for example, might require many receptive fields to move by large amounts to produce vision with high enough resolution.

Van Es et al. have now answered this question by using a brain scanner to measure receptive fields in healthy volunteers. The volunteers focused on different visual features, such as color or motion, and to various visual locations. When the volunteers attended to color, their attention was more tightly focused than when they attended to motion. This might be because processing color requires fine-detail vision, whereas we can detect movement with our attention spread over a larger area. As a result, receptive fields moved more when the volunteers attended to color than when they attended to motion.

Movement of visual receptive fields thus depends on what we attend to, as well as where we focus our attention. This adds to our understanding of how the brain filters the information bombarding our senses. This might lead to better diagnosis and treatment of disorders that include attentional problems, such as autism and ADHD. The results could also help develop artificial intelligence systems that, like the visual system, can process information flexibly to achieve different goals.

The resolution of the visual system is highest at the fovea and decreases gradually with increasing eccentricity. But the visual system’s resolution is not fixed. Attention can be directed to a location in space and/or a visual feature, which temporarily improves behavioral performance (Posner et al., 1980; Rossi and Paradiso, 1995; Found and Müller, 1996; Carrasco and Yeshurun, 1998; Yeshurun and Carrasco, 1999; Kumada, 2001; Sàenz et al., 2003; Wolfe et al., 2003; Theeuwes and Van der Burg, 2007) at the cost of reduced sensitivity for non-attended locations and features (Kastner and Pinsk, 2004; Pestilli and Carrasco, 2005; Wegener et al., 2008).

Attending a location in space increases activity in units representing the attended location, as shown by both electrophysiological (Luck et al., 1997; Reynolds et al., 2000) and fMRI studies (Tootell et al., 1998; Silver et al., 2005; Datta and DeYoe, 2009). In addition, spatial receptive fields were shown to shift toward an attended location in macaque MT+ (Womelsdorf et al., 2006) and V4 (Connor et al., 1997). Using fMRI to measure population receptive fields (pRFs; Dumoulin and Wandell, 2008; Dumoulin and Knapen, 2018), it was found that pRF shifts induced by spatial attention occur throughout human visual cortex (Klein et al., 2014; Kay et al., 2015; Sheremata and Silver, 2015; Vo et al., 2017), a process thought to improve visual resolution at the attended location (Anton-Erxleben and Carrasco, 2013; Kay et al., 2015; Vo et al., 2017). Such spatial resampling is understood to be the result of an interaction between bottom-up sensory signals and a top-down attentional gain field (Womelsdorf et al., 2008; Klein et al., 2014; Miconi and VanRullen, 2016).

Feature-based attention, for example directed toward color or motion, selectively increases activity in those units that represent the attended feature, as evidenced by electrophysiological (Treue and Maunsell, 1996; Treue et al., 1999; McAdams and Maunsell, 2000; Maunsell and Treue, 2006; Müller et al., 2006; Zhang and Luck, 2009; Zhou and Desimone, 2011), fMRI (Saenz et al., 2002; Serences and Boynton, 2007; Jehee et al., 2011), and behavioral reports (Sàenz et al., 2003; White and Carrasco, 2011). These studies consistently show that feature-based attention modulates processing irrespective of the attended stimulus's spatial location. In addition, feature-based attention also appears to shift featural tuning curves toward the attended value, as reported by both electrophysiological (Motter, 1994; David et al., 2008) and fMRI studies (Çukur et al., 2013).

The similarity in the effects of feature-based and spatial attention on affected neural units suggests a common neural mechanism for both sources of attention (Hayden and Gallant, 2005; Cohen and Maunsell, 2011). Yet spatial attention necessitates retinotopically precise feedback (Miconi and VanRullen, 2016), whereas feature-based attention operates throughout the visual field (Maunsell and Treue, 2006). Studies investigating whether one source of attention potentiates the other generally find that interactions are either nonexistent or very weak at the earliest stages of processing (David et al., 2008; Hayden and Gallant, 2009; Patzwahl and Treue, 2009; Zhang and Luck, 2009), but emerge at later stages of visual processing (Hillyard and Münte, 1984; Handy et al., 2001; Bengson et al., 2012; Ibos and Freedman, 2016), and ultimately influence behavior (Kingstone, 1992; Kravitz and Behrmann, 2011; Leonard et al., 2015; White et al., 2015; Nordfang et al., 2018). In addition, the effects of feature-based compared to spatial attention arise earlier in time. This occurs both when only feature-based attention is endogenously cued and subsequently guides spatial attention towards the attended feature's location (Hopf et al., 2004), and when both types of attention are endogenously cued (Hayden and Gallant, 2005; Andersen et al., 2011). This supports the idea that feature-based attention can direct spatial attention toward or away from specific locations containing attended or unattended features (Cohen and Shoup, 1997; Cepeda et al., 1998; Burnett et al., 2016).

The studies mentioned above investigated modulatory effects of feature-based attention on spatial attention by measuring changes in response amplitude (e.g. ERP/firing rate). However, no study to date has investigated the effect of feature-based attention on spatial sampling. Yet, exactly this relationship is predicted by behavioral studies. Especially when attention is endogenously cued, feature-based attention has been argued to influence the spatial resampling induced by spatial attention to optimize sampling of visual features for behavior (Yeshurun and Carrasco, 1998; Yeshurun and Carrasco, 2000; Yeshurun et al., 2008; Barbot and Carrasco, 2017). Specifically, these authors suggested that when attending features that are processed by neurons with smaller receptive fields, a greater degree of spatial resampling is required to resolve the required featural resolution at the attended location. In the current study, we put this hypothesis to the test by measuring the brain's representation of space under conditions of differential attention.

Specifically, we measured pRFs under conditions of differential spatial attention (i.e. toward fixation or the mapping stimulus) and feature-based attention (i.e. toward the mapping stimulus's temporal frequency or color content). One important reason for studying the effects of attention to color and temporal frequency is that they are known to be processed at different spatial scales. Specifically, color is generally processed at a finer spatial scale compared to temporal frequency. First, this is a result of color compared to temporal frequency information being processed predominantly by the parvocellular rather than the magnocellular pathways, which in turn pool differentially across visual space (Schiller and Malpeli, 1978; Hicks et al., 1983; Denison et al., 2014). Second, temporal frequency and color are processed across different cortical areas (i.e. MT +compared to hV4; Liu and Wandell (2005); Brouwer and Heeger, 2009, Brouwer and Heeger, 2013; Winawer et al., 2010) that have differential spatial precision (Amano et al., 2009; Winawer et al., 2010). Third, preference for color compared to temporal frequency is generally greater in the fovea compared to the periphery (Curcio et al., 1990; Azzopardi et al., 1999; Brewer et al., 2005), where receptive fields are generally smaller (Dumoulin and Wandell, 2008). The hypothesized relation between spatial scale of attended features and spatial resampling as proposed by the behavioral studies mentioned above thus implies that attending color compared to temporal frequency at a particular location should lead to stronger spatial resampling. In addition, as color and temporal frequency are differentially processed across cortical areas, studying these features allows us to investigate whether modulations of spatial resampling by feature-based attention are specific for areas that prefer the attended feature. We specifically chose temporal frequency and not coherent motion, as coherent motion signals have been shown to influence pRF measurements (Harvey and Dumoulin, 2016). It was previously shown that attention can be directed to both feature domains (Wolfe and Horowitz, 2004; Cass et al., 2011).

We characterized how spatial attention influences the sampling of visual space, and subsequently investigated how feature-based attention modulates this spatial resampling. In addition, an explicit gain-field interaction model allowed us to formally capture the pRF position changes resulting from our attentional manipulations (Klein et al., 2014). Finally, we also performed a full-field stimulation experiment which allowed us to relate these attentional modulations to each voxel's bottom-up preference for color and temporal frequency.

In brief, our results show that pRF changes are indeed stronger when attending the stimulus's color compared to temporal frequency content. These modulations occurred similarly throughout the visual system, regardless of an area's bottom-up feature preference. We show that these feature-based attentional modulations can be explained by changes in the precision of the attentional gain field. Together, this confirms the idea that the degree of spatial resampling is dependent on the spatial scale at which attended features are processed.

For the purpose of our study, we adapted a standard traversing bar-stimulus based retinotopic mapping paradigm (Dumoulin and Wandell, 2008). Specifically, our bar stimulus was made up out of many Gabor elements, each with one of two color compositions (blue/yellow or green/magenta) and with particular temporal frequency (high or low). Throughout the duration of a bar pass, participants reported changes in fixation mark luminance, in the predominant Gabor color composition, or in the predominant Gabor temporal frequency (see Figure 1A and Materials and methods section for additional details). Importantly, visual stimulation was identical across conditions and only the top-down factor of attention was manipulated. In addition, behavioral difficulty was kept constant between conditions and across three levels of eccentricity using a Quest staircase procedure. Thus, spatial attention was either directed towards the fixation mark (Attend Fixation) or towards the bar stimulus (Attend Stimulus). In addition, feature-based attention was either directed towards color (Attend Color), or towards temporal frequency (Attend TF). This setup allowed us to fit separate population receptive field (pRF) models to data from the different attentional conditions for each voxel (see Figure 1B and Materials and methods section for additional details).

Figure 1

Download asset Open asset

Our principal aim is to understand whether feature-based attention influences spatial resampling induced by spatial attention. Therefore, we first characterize in detail the changes in spatial sampling (i.e. pRF properties) that resulted from differential allocation of spatial attention. Second, we capture these pRF modulations by spatial attention in an attentional gain field modeling framework. Third, we investigate how feature-based attention modulated the pattern of pRF changes and how it affected attentional gain field model parameters. Finally, we relate these feature-based attentional modulations to (1) bottom-up feature preference and (2) to differences in the spatial scale at which color and temporal frequency are processed.

Region of interest definition

Figure 2A shows voxels' Attend Fixation location preferences, by depicting color-coded polar angle coordinates on an inflated cortical surface for one example participant's right hemisphere. We examined the relation between pRF eccentricity and size within each of the retinotopic regions, and performed further analyses on those regions that showed clear progressions of polar angle on the surface as well as positive size-eccentricity relations, as shown in Figure 2B. In addition, we created a combined ROI that pooled voxels across selected ROIs to evaluate pRF changes across the visual system.

Figure 2 with 2 supplements see all

Download asset Open asset

pRF changes induced by spatial attention

To quantify pRF changes resulting from differential allocation of spatial attention, we created an Attend Stimulus condition by averaging pRF parameters between the Attend Color and Attend TF conditions. To inspect how spatial attention affected pRF positions, we plotted a vector from the Attend Fixation to the Attend Stimulus pRF position. For visualization purposes, we created visual field quadrant representations by multiplying both the Attend Fixation and Attend Stimulus pRF x- and y-coordinates with the sign of pRF x- and y-coordinates in the Attend Fixation condition (see Figure 3A). This means, for example, that pRFs in the upper-right quadrant were unaffected (i.e. x- and y-coordinate multiplied by 1), while pRFs in the lower-right quadrant were mirrored along the y-axis (i.e. x-coordinate multiplied by 1, y-coordinate multiplied by −1). Note that mirroring based on the Attend Fixation condition preserves any pRF shifts across a meridian, allowing pRFs to shift outside the target visual field quadrant. Visual inspection of these pRF position shifts shows both increasing shift magnitude up the visual hierarchy and shifts occurring mainly along the radial dimension (i.e. toward or away from the fovea; Figure 3B).

Figure 3 with 3 supplements see all

Download asset Open asset

pRF changes across eccentricity

To further inspect the attention-induced radial shifts described above, we plotted the difference between Attend Stimulus and Attend Fixation pRF eccentricity for each of four Attend Fixation pRF eccentricity bins (Figure 4A). This returned varying patterns of pRF eccentricity changes across the different ROIs. The combined ROI shows that overall, parafoveal pRFs shifted away from the fovea, while peripheral pRFs shifted toward the fovea. These outward shifting parafoveal pRFs are found in all other ROIs except V1 and V2, whereas the inward shifting peripheral pRFs are also present in V1, V2 and V3 (see Supplementary file 1 -Tables 4 and 5).

Figure 4 with 2 supplements see all

Download asset Open asset

In addition to pRF position changes, we also inspected changes in pRF size induced by differences in spatial attention as a function of Attend Fixation pRF eccentricity (Figure 4B). Overall, parafoveal pRFs increased in size, while peripheral pRFs decreased in size. These expanding parafoveal pRFs were present in all ROIs except V2/3, whereas shrinking peripheral pRFs were found in all ROIs except V1, MT+ and IPS0 (see Supplementary file 1 -Tables 6 and 7). Overall, this pattern of results is strikingly similar to the changes in pRF eccentricity described above. In fact, the changes in pRF size and eccentricity were strongly correlated in all ROIs (Figure 4C, Pearson R over 20 5-percentile bins between 0.74 and 0.99, p's < 0.001, see Supplementary file 1 -Table 8). Together, these results show that attention to the stimulus caused parafoveal pRFs to shift away from the fovea and increase in size, whereas peripheral pRFs shifted toward the fovea and decreased in size.

Formal account for observed pattern of pRF shifts

To provide a mechanistic explanation for the complex pattern of pRF shifts described above, we modeled our results using a multiplicative Gaussian gain field model (Womelsdorf et al., 2008; Klein et al., 2014). We adapted this framework to work in conditions where attention shifted over space as a function of time (see Materials and methods). In brief, this modeling procedure used the Attend Fixation pRF, one attentional gain field at fixation and another convolved with the stimulus to predict the Attend Stimulus pRF position. We determined optimal attentional gain field sizes by minimizing the difference between observed and predicted Attend Stimulus pRF positions in the quadrant visual field format of Figure 3B. Figure 5A illustrates that model predictions closely followed the data, thereby accurately reproducing radially shifting pRFs. Examining the predicted change in pRF eccentricity as a function of eccentricity (i.e. the dominant pRF shift direction; Figure 5B) showed that the model was able to capture widely varying eccentricity change profiles across ROIs using very similar attentional gain field sizes (Figure 5C). This shows that a common attentional influence can result in very different pRF shift patterns, which then depend on differential spatial sampling properties across ROIs (i.e. distribution of pRF sizes and positions). In sum, these results show that the attentional gain field model provides a parsimonious and powerful account for the variety of pRF shift patterns across ROIs.

Figure 5 with 2 supplements see all

Download asset Open asset

We further investigated how the model was able to reproduce the eccentricity-dependent eccentricity changes we reported above. For this, we inspected pRF shifts induced by attending either fixation or the stimulus relative to the stimulus drive (i.e. the pRF outside the influence of attention derived from the model). For illustrative purposes, we display results for V2, V3 and IPS0 as these areas showed marked differences in their eccentricity change profile (Figure 5D–F). The left panels of each figure reveal the effects of attending fixation and the stimulus separately. This shows that both sources of spatial attention pull the measured pRFs toward the fovea, albeit with differing relative magnitudes across eccentricity. The right panel of each figure shows that the resulting difference between attending fixation and the stimulus constitutes the eccentricity-dependent patterns observed in the data (Figure 5B).

Together, these analyses show that existing multiplicative gain field models of attention can be extended to predict pRF shifts in situations where spatial attention shifts over time. Additionally, it confirms, extends and quantifies earlier reports showing that the precision of the attentional gain field is similar across the visual hierarchy (Klein et al., 2014).

Feature-based attentional modulation

Having established (1) the pattern of changes in spatial sampling (i.e. changes in pRF size and eccentricity) resulting from differential allocation of spatial attention, and (2) a mechanistic explanation of these changes, we next examined how this pattern was modulated by differences in feature-based attention. Figure 6A shows how pRF eccentricity and size are differentially affected by attending color or temporal frequency within the stimulus for the combined ROI. This illustrates that while both tasks caused similar pRF changes, these effects were generally more pronounced when attending color.

Figure 6 with 4 supplements see all

Download asset Open asset

To quantify the modulation of feature-based attention per voxel, we first set up a single robust index of the degree to which spatial attention resampled visual space, combining changes in pRF eccentricity and size (as these were highly correlated, see Figure 4C). This Attentional Modulation Index (AMI, see Materials and methods) is depicted in Figure 6B for the combined ROI when attending color and TF. We then quantified the difference in this AMI between attending color and temporal frequency as a feature-based Attentional Modulation Index (feature AMI, see Materials and methods). Positive values of feature AMI indicate that attending color induced greater pRF changes, while negative values indicate that attending TF led to stronger pRF changes. Figure 6C shows that this feature AMI was positive across eccentricity in the combined ROI. Inspecting the average feature AMI across voxels within each ROI (Figure 6D) reveals that attending changes in color compared to TF in the bar stimulus produced stronger spatial resampling in all ROIs (p's < 0.01, see Supplementary file 1 -Table 9). Specifically, the feature AMI was around 0.05 on average across ROIs. As the feature AMI is a contrast measure where difference is divided by the sum, this corresponds to roughly 10% stronger pRF changes when attending color compared to temporal frequency. In some ROIs (V3AB/IPS0, see Figure 6—figure supplement 1), the FAMI reached values of 0.20, which corresponds roughly to 50% stronger pRF changes when attending color compared to temporal frequency. Computing the AMI with either pRF eccentricity or size changes separately (i.e. not as a combined measure) yields similar results (see Figure 6—figure supplement 2).

Feature preference and spatial sampling

How do we explain that attending color in the stimulus induced greater changes in spatial sampling? Behavioral studies have suggested that the influence of spatial attention should be adjusted by feature-based attention to improve sampling of attended visual features (Yeshurun et al., 2008; Barbot and Carrasco, 2017). Specifically, these authors suggested that attending features processed by relatively smaller receptive fields requires a greater degree of spatial resampling. This implies that if color-preferring voxels are relatively small, this could explain the greater degree of resampling observed when attending color. This is indeed predicted by the fact that color compared to temporal frequency information is predominantly processed by the parvocellular compared to the magnocellular pathway, where spatial sampling is generally more fine-grained (Schiller and Malpeli, 1978; Hicks et al., 1983; Denison et al., 2014). In addition, pRF size varies on average between visual regions, such that pRF size is generally larger in area MT+ (preferring temporal frequency) compared to hV4 (preferring color). Finally, both pRF size (Dumoulin and Wandell, 2008) and color compared to TF preference (Curcio et al., 1990; Azzopardi et al., 1999; Brewer et al., 2005) are known to be strongly eccentricity-dependent such that foveal voxels have relatively small pRFs and are relatively color-sensitive. We also clearly observe both effects in our data (see Figure 2B and Figure 7, correlation between feature-preference and eccentricity is negative except in LO and VO, see Supplementary file 1 - Table 10). In sum, the greater amount of spatial resampling when attending color can be parsimoniously explained by color being sampled by relatively smaller pRFs.

Figure 7 with 1 supplement see all

Download asset Open asset

Feature-based attention influences attentional gain field precision

Smaller pRFs also require a more precise attentional gain field to shift a given distance (a property of the multiplication of Gaussians). Combining this with our observation that pRFs experience greater shifts when attending color, we predict that attentional gain fields should be more precise in this condition. To test this, we repeated the attentional gain field modeling procedure described above, replacing the Attend Stimulus data with the Attend Color and Attend TF data in two separate fit procedures. Both our data (Figure 5) and previous findings (Klein et al., 2014) showed that a single attentional gain field affects the different visual regions similarly. In addition, our results presented above showed that pRF modulation by feature-based attention was not related to feature preference across ROIs. We therefore analyzed feature-based attentional modulations of the attentional gain field both on data from all ROIs fitted together (the 'combined ROI'), and as the median across individually fitted ROIs. This analysis returned smaller fitted stimulus attentional gain field sizes in the Attend Color compared to the Attend TF fit procedure (Figure 8) both in the combined ROI (0.094 dva smaller over subjects when attending color, t(4) = 9.021, p = 0.001, Cohen's d = 4.511) and across ROIs (median over ROIs on average 0.061 dva smaller over subjects when attending color, t(4) = 4.243, p = 0.013, Cohen's d = 2.121). As the Attend Fixation data were used as input in both modeling procedures, we verified that the estimated fixation attentional gain field was not different between procedures (p's of. 693 and. 224 and Cohen's d of −0.213 and −0.719 for across ROIs and combined ROI, respectively). These analyses show that the stronger influence of spatial attention when attending color is realized by a more precise attentional gain field located at the stimulus. In sum, our results suggest that (1) the attentional system adjusts its influence in accordance with the spatial sampling characteristics of units that prefer the attended feature and (2) that it does this equally across visual regions regardless of their bottom-up feature preference.

Figure 8

Download asset Open asset

Task and fixation performance

Finally, we verified that the pRF results were not affected by differences in fixation accuracy or behavioral performance (see Figure 9). To provide evidence in favor of these null hypotheses, we performed JZL Bayes factor analyses (using JASP; Love et al., 2015) as frequentist statistics are not capable of providing evidence for the null (Altman and Bland, 1995). We rotated recorded eye position to the direction of bar movement and computed the median and standard deviation of position along this dimension across bar passes per bar position (Figure 9B). We next set up a model including the factor of attention condition (3 levels), bar position (24 levels) and their interaction. We found that when predicting gaze position, the evidence was in favor of the null hypothesis with a Bayes Factor (BF) of 18620. When predicting gaze variability, however, we found evidence against the null hypothesis with a BF of 5.980. Evidence for including each of the factors (condition, bar position and their interaction) into the model returned BFs of 0.713, 547.193 and 0.017, respectively. The BF of 0.713 for the factor of condition means that we cannot determine whether gaze variability was different between conditions. However, even if this were the case, this could only lead to an offset in pRF size and not to changes in pRF position (Levin et al., 2010; Klein et al., 2014; Hummer et al., 2016). In addition, any anisotropy in gaze position variability could potentially lead to offsets in pRF center positions. Nevertheless, these biases would be identical for all pRFs throughout the visual field. As we find that peripheral pRFs shift inwards and decrease in size and central pRFs shift outwards and increase in size, global offsets in pRF size and position cannot parsimoniously explain our results. More importantly, the analyses also showed that although bar position influenced gaze variability (BF of 547.193), it did not do so differently between attention conditions (BF of 0.017).

Figure 9 with 1 supplement see all

Download asset Open asset

Although we used a Quest procedure to equate difficulty across attention conditions and across different levels of eccentricity, it is possible that this procedure stabilized at a faulty difficulty level. To verify whether the Quest procedure successfully equated performance, we used a similar Bayesian approach, testing whether a model including attention condition (three levels) and stimulus eccentricity (three levels) influenced behavioral performance (Figure 9A). This returned evidence for the null hypothesis with a BF of 6.25. Together, these results show that differences in pRF parameters between conditions cannot be explained by either fixation accuracy or behavioral difficulty.

We investigated how spatial and feature-based attention jointly modulate the sampling of visual space. We found that directing covert spatial attention toward a moving bar stimulus altered the eccentricity and size of pRFs in concert. These changes in spatial sampling were parsimoniously explained by an attentional gain field model. Attending color changes within this stimulus induced stronger pRF changes compared to attending temporal frequency changes. These feature-based attentional modulations occurred globally throughout the brain, irrespective of a visual region's average feature preference. We suggest that the greater degree of spatial resampling when attending color is related to smaller pRF sizes in relatively color preferring voxels. In addition, we showed that the greater degree of spatial resampling when attending color is caused by a more precise attentional gain field on the stimulus.

Previous behavioral reports suggested that the spatial scale at which an attended feature is processed influenced the degree of spatial resampling (Yeshurun and Carrasco, 1998; Yeshurun and Carrasco, 2000; Yeshurun et al., 2008; Barbot and Carrasco, 2017). Specifically, features processed at a finer spatial scale require a greater degree of spatial resampling. This means that the greater degree of spatial resampling when attending color compared to temporal frequency could be explained by color being sampled at a finer spatial scale. This is a canonical difference between parvocellular small, color-sensitive receptive fields and magnocellular large, temporal frequency-sensitive receptive fields (Schiller and Malpeli, 1978; Hicks et al., 1983; Denison et al., 2014). In addition, both our data and previous findings show that color is preferentially processed by visual areas that on average have smaller receptive field sizes (i.e. hV4 compared to MT+ in Liu and Wandell (2005); Amano et al., 2009; Brouwer and Heeger, 2009, Brouwer and Heeger, 2013; Winawer et al., 2010). Finally, both the current and previous studies show that pRF size (Dumoulin and Wandell, 2008) and color compared to temporal frequency preference (Curcio et al., 1990; Azzopardi et al., 1999; Brewer et al., 2005) vary across eccentricity such that foveal voxels have smaller pRFs and are more color-sensitive. In sum, we suggest that the greater degree of spatial resampling when attending color compared to temporal frequency can be explained by the difference in spatial scale at which these features are processed. We therefore predict that our results should generalize to any other comparison of attended visual features as long as these features differ in their spatial scale. This includes attending different feature values such as high compared to low spatial frequency, or attending different feature dimensions such as faces (broader spatial scale) versus letters (finer spatial scale).

Electrophysiological studies on the interaction between feature-based and spatial attention generally measure overall response amplitudes rather than changes in spatial sampling. This implies that interactions between feature-based and spatial attention are weak to non-existent in relatively early stages of processing (David et al., 2008; Hayden and Gallant, 2009; Patzwahl and Treue, 2009; Zhang and Luck, 2009), but develop at later stages of visual processing (Hillyard and Münte, 1984; Handy et al., 2001; Andersen et al., 2011; Bengson et al., 2012; Ibos and Freedman, 2016), but see Egner et al., 2008). We add to this (1) that feature-based attention modulates the effects of spatial attention on spatial resampling, and (2) that these interactions occur globally throughout the brain, manifesting themselves in even the earliest cortical visual regions. Interactions between spatial and feature-based attention in the early stages of processing could be concealed when focusing on changes in response amplitude rather than changes in spatial sampling. However, it is important to note that measuring spatial sampling at the level of voxels does not allow us to determine whether observed changes in spatial sampling are the result of changes in spatial sampling of individual neurons, or rather the result of differential weighting of subpopulations of neurons within a voxel. Nevertheless, it has been shown that spatial attention does influence spatial sampling of individual neurons (Connor et al., 1997; Womelsdorf et al., 2006). Yet, future studies are required to extend our conclusions regarding the interactions between feature-based and spatial attention to the single neuron level.

The greater degree of spatial resampling when attending color compared to temporal frequency occurred throughout the brain, irrespective of visual regions' preference for the attended features. In other words, while MT+ and hV4 differed greatly in their relative feature preference, both areas showed comparable pRF changes resulting from differences in feature-based attention. This stands in apparent contrast to previous studies reporting that feature-based attention selectively enhances responses in cortical areas specialized in processing the attended feature (Corbetta et al., 1990; Chawla et al., 1999; O'Craven et al., 1999; Schoenfeld et al., 2007; Baldauf and Desimone, 2014). However, attending a stimulus consisting of multiple features was shown to spread attentional response modulations of one of the object's feature dimensions to other constituent feature dimensions (Katzner et al., 2009; Çukur et al., 2013; Kay and Yeatman, 2017), albeit somewhat later in time (±60 ms; Schoenfeld et al., 2014). This could mean that the global pattern of pRF shifts we observed is caused by such an object-based attentional transfer mechanism. In addition, changing the sampling of visual space globally throughout the brain enhances stability in the representation of space. Different modifications of visual space per visual region would require an additional mechanism linking different spatial representations. Instead, the global nature of spatial resampling we observe supports a temporally dynamic but spatially consistent adaptation of visual space.

An important remaining question pertains to the source of the interactions between feature-based and spatial attention. Signals of spatial selection are thought to originate from a network of frontal and parietal areas, identified using fMRI (Shulman et al., 2002; Silver et al., 2005; Jerde et al., 2012; Sprague and Serences, 2013; Szczepanski et al., 2013; Kay and Yeatman, 2017; Mackey et al., 2017) and electrophysiology (Moore and Armstrong, 2003; Gregoriou et al., 2009). As we focused on careful measurement of spatial sampling in feature-sensitive visual cortex with a relatively small stimulus region, we did not include the frontoparietal regions containing large receptive fields in our analyses. A recent study suggested a central role for the ventral prearcuate gyrus for conjoined spatial and feature-based attentional modulations (Bichot et al., 2015). Correspondingly, signals of feature selection in humans have been localized to a likely human homologue of this area, the inferior frontal junction (IFJ; Zanto et al., 2010; Baldauf and Desimone, 2014). This region is therefore a possible candidate for controlling the interactions between feature-based and spatial attention.

The average changes in pRF size and eccentricity for each visual region in our data are largely consistent with previous studies in which attention was devoted to a peripheral stimulus versus fixation (Kay et al., 2015; Sheremata and Silver, 2015). Moreover, our analyses go beyond these average pRF changes by investigating the spatial structure of the complex pattern of pRF changes that resulted from such differential spatial attention. The resulting characterization details how the sampling of visual space by single voxel pRFs is affected by spatial attention, which is of specific relevance for future studies that determine spatial selectivity for voxel selections. First, we show that attending the stimulus compared to fixation caused pRFs to shift radially. Although a previous study reported a dominant horizontal shift direction (Sheremata and Silver, 2015), we suggest that the overrepresentation of the horizontal meridian (Schneider et al., 2004; Swisher et al., 2007) made radially shifting pRFs appear as predominantly horizontal changes. Second, we report closely coupled pRF eccentricity and size changes that were dependent on pRF eccentricity. Specifically, we found that parafoveal pRFs shifted toward the periphery and increased in size, whereas peripheral pRFs shifted toward the fovea and decreased in size. This finding supports the resolution hypothesis of attention (Anton-Erxleben and Carrasco, 2013), which posits that spatial attention acts to reduce resolution differences between the fovea and periphery. We note that the functional implication of pRF size changes was recently questioned, as stimulus encoding fidelity was shown to be unaffected by pRF size changes (Vo et al., 2017). However, the exact functional significance of changes in pRF size bears no consequence for the conclusions currently presented. As we observed a strong correlation between pRF eccentricity and size changes, we combined both measures into a single robust index. Our relevant quantifications are therefore agnostic to the potentially separable functional implications of changes in pRF size and eccentricity.

The pattern of pRF shifts we observe is described well by an attentional gain field model (Reynolds and Heeger, 2009; Klein et al., 2014). First, this highlights that a simple and well-understood mechanism underpins the apparent complexity of the observed pattern of pRF changes. Second, it extends the utility of such attentional gain field models to situations in which attention is dynamically deployed over space and time during the mapping of the pRF (Kay et al., 2015). In agreement with earlier reports (Klein et al., 2014; Puckett and DeYoe, 2015), we found that the best-fitting model implemented comparable attentional gain field sizes across visual regions. This strongly points to spatial attention being implemented as a global influence across visual cortex. We conclude that differences in pRF shift patterns between different visual regions depended primarily on differences in visual sampling (i.e. differences in pRF center and size distributions) rather than on differing attentional influences. Despite the broad correspondence between model fits and data, the model did not capture the observed decreases in pRF eccentricity of the most foveal pRFs in V1. Two recent studies showed that in early visual areas, spatial attention shifted pRFs away from the attended location, but toward the attended location in higher visual areas (de Haas et al., 2014; Vo et al., 2017). Other studies showed that in precisely these visual regions, both the pRF and the attentional gain field are composed of a suppressive surround in addition to their positive peak (Zuiderbaan et al., 2012; Puckett and DeYoe, 2015). We leave the question of whether these suppressive surrounds could explain such repulsive shifts in lower visual cortex for future research. As a more general aside, gain fields have been shown to influence visual processing at the motor stage (Van Opstal et al., 1995; Snyder et al., 1998; Trotter and Celebrini, 1999). Thus, our results further establish the close link between attentional and motor processes (Rizzolatti et al., 1987; Corbetta et al., 1998).

In sum, we show that visuospatial sampling is not only affected by attended locations but also depends on the spatial sampling properties of units that prefer attended visual features. The global nature of these modulations highlights the flexibility of the brain’s encoding of sensory information to meet task demands (Rosenholtz, 2016).

All attention-pRF and feature-GLM results have been deposited on Figshare at https://doi.org/10.6084/m9.figshare.c.4012717.v1, under a Public Domain Dedication License. pRF fitting and (statistical) parameter processing code is available at https://github.com/daanvanes/PRF_attention_analysis (copy archived at https://github.com/eLifeProduction/pRF_attention_analysis).

1. 1. Daniel Marten van Es
   2. Jan Theeuwes
   3. Tomas Knapen

   (2018) figshare

   pRF spatial and feature-based attention data.

   https://doi.org/10.6084/m9.figshare.c.4012717.v1

1. 1. Adams DL
   2. Horton JC

   (2003) A precise retinotopic map of primate striate cortex generated from the representation of angioscotomas

   The Journal of Neuroscience 23:3771–3789.

   https://doi.org/10.1523/JNEUROSCI.23-09-03771.2003

   • PubMed
   • Google Scholar
2. 1. Altman DG
   2. Bland JM

   (1995) Statistics notes: Absence of evidence is not evidence of absence

   Bmj 311:485.

   https://doi.org/10.1136/bmj.311.7003.485

   • Google Scholar
3. 1. Amano K
   2. Wandell BA
   3. Dumoulin SO

   (2009) Visual field maps, population receptive field sizes, and visual field coverage in the human MT+ complex

   Journal of Neurophysiology 102:2704–2718.

   https://doi.org/10.1152/jn.00102.2009

   • PubMed
   • Google Scholar
4. 1. Andersen SK
   2. Fuchs S
   3. Müller MM

   (2011) Effects of feature-selective and spatial attention at different stages of visual processing

   Journal of Cognitive Neuroscience 23:238–246.

   https://doi.org/10.1162/jocn.2009.21328

   • PubMed
   • Google Scholar
5. 1. Anton-Erxleben K
   2. Carrasco M

   (2013) Attentional enhancement of spatial resolution: linking behavioural and neurophysiological evidence

   Nature Reviews Neuroscience 14:188–200.

   https://doi.org/10.1038/nrn3443

   • PubMed
   • Google Scholar
6. 1. Azzopardi P
   2. Jones KE
   3. Cowey A

   (1999) Uneven mapping of magnocellular and parvocellular projections from the lateral geniculate nucleus to the striate cortex in the macaque monkey

   Vision Research 39:2179–2189.

   https://doi.org/10.1016/S0042-6989(98)00319-8

   • PubMed
   • Google Scholar
7. 1. Baldauf D
   2. Desimone R

   (2014) Neural mechanisms of object-based attention

   Science 344:424–427.

   https://doi.org/10.1126/science.1247003

   • PubMed
   • Google Scholar
8. 1. Barbot A
   2. Carrasco M

   (2017) Attention Modifies Spatial Resolution According to Task Demands

   Psychological Science 28:285–296.

   https://doi.org/10.1177/0956797616679634

   • PubMed
   • Google Scholar
9. 1. Bengson JJ
   2. Lopez-Calderon J
   3. Mangun GR

   (2012) The spotlight of attention illuminates failed feature-based expectancies

   Psychophysiology 49:1101–1108.

   https://doi.org/10.1111/j.1469-8986.2012.01385.x

   • PubMed
   • Google Scholar
10. 1. Bichot NP
    2. Heard MT
    3. DeGennaro EM
    4. Desimone R

    (2015) A Source for Feature-Based Attention in the Prefrontal Cortex

    Neuron 88:832–844.

    https://doi.org/10.1016/j.neuron.2015.10.001

    • PubMed
    • Google Scholar
11. 1. Brewer AA
    2. Liu J
    3. Wade AR
    4. Wandell BA

    (2005) Visual field maps and stimulus selectivity in human ventral occipital cortex

    Nature Neuroscience 8:1102–1109.

    https://doi.org/10.1038/nn1507

    • PubMed
    • Google Scholar
12. 1. Brouwer GJ
    2. Heeger DJ

    (2009) Decoding and reconstructing color from responses in human visual cortex

    Journal of Neuroscience 29:13992–14003.

    https://doi.org/10.1523/JNEUROSCI.3577-09.2009

    • PubMed
    • Google Scholar
13. 1. Brouwer GJ
    2. Heeger DJ

    (2013) Categorical clustering of the neural representation of color

    Journal of Neuroscience 33:15454–15465.

    https://doi.org/10.1523/JNEUROSCI.2472-13.2013

    • PubMed
    • Google Scholar
14. 1. Buracas GT
    2. Boynton GM

    (2002) Efficient design of event-related fMRI experiments using M-sequences

    NeuroImage 16:801–813.

    https://doi.org/10.1006/nimg.2002.1116

    • PubMed
    • Google Scholar
15. 1. Burnett KE
    2. Close AC
    3. d'Avossa G
    4. Sapir A

    (2016) Spatial attention can be biased towards an expected dimension

    Quarterly Journal of Experimental Psychology 69:2218–2232.

    https://doi.org/10.1080/17470218.2015.1111916

    • PubMed
    • Google Scholar
16. 1. Carrasco M
    2. Yeshurun Y

    (1998) The contribution of covert attention to the set-size and eccentricity effects in visual search

    Journal of Experimental Psychology: Human Perception and Performance 24:673–692.

    https://doi.org/10.1037/0096-1523.24.2.673

    • Google Scholar
17. 1. Cass J
    2. Van der Burg E
    3. Alais D

    (2011) Finding flicker: critical differences in temporal frequency capture attention

    Frontiers in Psychology 2:320.

    https://doi.org/10.3389/fpsyg.2011.00320

    • PubMed
    • Google Scholar
18. 1. Cepeda NJ
    2. Cave KR
    3. Bichot NP
    4. Kim MS

    (1998) Spatial selection via feature-driven inhibition of distractor locations

    Perception & Psychophysics 60:727–746.

    https://doi.org/10.3758/BF03206059

    • PubMed
    • Google Scholar
19. 1. Chawla D
    2. Rees G
    3. Friston KJ

    (1999) The physiological basis of attentional modulation in extrastriate visual areas

    Nature Neuroscience 2:671–676.

    https://doi.org/10.1038/10230

    • PubMed
    • Google Scholar
20. 1. Cohen MR
    2. Maunsell JH

    (2011) Using neuronal populations to study the mechanisms underlying spatial and feature attention

    Neuron 70:1192–1204.

    https://doi.org/10.1016/j.neuron.2011.04.029

    • PubMed
    • Google Scholar
21. 1. Cohen A
    2. Shoup R

    (1997) Perceptual dimensional constraints in response selection processes

    Cognitive Psychology 32:128–181.

    https://doi.org/10.1006/cogp.1997.0648

    • PubMed
    • Google Scholar
22. 1. Connor CE
    2. Preddie DC
    3. Gallant JL
    4. Van Essen DC

    (1997) Spatial attention effects in macaque area V4

    The Journal of Neuroscience 17:3201–3214.

    https://doi.org/10.1523/JNEUROSCI.17-09-03201.1997

    • PubMed
    • Google Scholar
23. 1. Corbetta M
    2. Miezin FM
    3. Dobmeyer S
    4. Shulman GL
    5. Petersen SE

    (1990) Attentional modulation of neural processing of shape, color, and velocity in humans

    Science 248:1556–1559.

    https://doi.org/10.1126/science.2360050

    • PubMed
    • Google Scholar
24. 1. Corbetta M
    2. Akbudak E
    3. Conturo TE
    4. Snyder AZ
    5. Ollinger JM
    6. Drury HA
    7. Linenweber MR
    8. Petersen SE
    9. Raichle ME
    10. Van Essen DC
    11. Shulman GL

    (1998) A common network of functional areas for attention and eye movements

    Neuron 21:761–773.

    https://doi.org/10.1016/S0896-6273(00)80593-0

    • PubMed
    • Google Scholar
25. 1. Çukur T
    2. Nishimoto S
    3. Huth AG
    4. Gallant JL

    (2013) Attention during natural vision warps semantic representation across the human brain

    Nature Neuroscience 16:763–770.

    https://doi.org/10.1038/nn.3381

    • PubMed
    • Google Scholar
26. 1. Curcio CA
    2. Sloan KR
    3. Kalina RE
    4. Hendrickson AE

    (1990) Human photoreceptor topography

    The Journal of Comparative Neurology 292:497–523.

    https://doi.org/10.1002/cne.902920402

    • PubMed
    • Google Scholar
27. 1. Datta R
    2. DeYoe EA

    (2009) I know where you are secretly attending! The topography of human visual attention revealed with fMRI

    Vision Research 49:1037–1044.

    https://doi.org/10.1016/j.visres.2009.01.014

    • PubMed
    • Google Scholar
28. 1. David SV
    2. Hayden BY
    3. Mazer JA
    4. Gallant JL

    (2008) Attention to stimulus features shifts spectral tuning of V4 neurons during natural vision

    Neuron 59:509–521.

    https://doi.org/10.1016/j.neuron.2008.07.001

    • PubMed
    • Google Scholar
29. 1. de Haas B
    2. Schwarzkopf DS
    3. Anderson EJ
    4. Rees G

    (2014) Perceptual load affects spatial tuning of neuronal populations in human early visual cortex

    Current Biology 24:R66–R67.

    https://doi.org/10.1016/j.cub.2013.11.061

    • PubMed
    • Google Scholar
30. 1. Denison RN
    2. Vu AT
    3. Yacoub E
    4. Feinberg DA
    5. Silver MA

    (2014) Functional mapping of the magnocellular and parvocellular subdivisions of human LGN

    NeuroImage 102:358–369.

    https://doi.org/10.1016/j.neuroimage.2014.07.019

    • PubMed
    • Google Scholar
31. 1. Dumoulin SO
    2. Knapen T

    (2018) How visual cortical organization is altered by ophthalmologic and neurologic disorders

    Annual Review of Vision Science 4:357–379.

    https://doi.org/10.1146/annurev-vision-091517-033948

    • PubMed
    • Google Scholar
32. 1. Dumoulin SO
    2. Wandell BA

    (2008) Population receptive field estimates in human visual cortex

    NeuroImage 39:647–660.

    https://doi.org/10.1016/j.neuroimage.2007.09.034

    • PubMed
    • Google Scholar
33. 1. Egner T
    2. Monti JM
    3. Trittschuh EH
    4. Wieneke CA
    5. Hirsch J
    6. Mesulam MM

    (2008) Neural integration of top-down spatial and feature-based information in visual search

    Journal of Neuroscience 28:6141–6151.

    https://doi.org/10.1523/JNEUROSCI.1262-08.2008

    • PubMed
    • Google Scholar
34. 1. Found A
    2. Müller HJ

    (1996) Searching for unknown feature targets on more than one dimension: investigating a "dimension-weighting" account

    Perception & Psychophysics 58:88–101.

    https://doi.org/10.3758/BF03205479

    • PubMed
    • Google Scholar
35. 1. Gregoriou GG
    2. Gotts SJ
    3. Zhou H
    4. Desimone R

    (2009) High-frequency, long-range coupling between prefrontal and visual cortex during attention

    Science 324:1207–1210.

    https://doi.org/10.1126/science.1171402

    • PubMed
    • Google Scholar
36. 1. Handy TC
    2. Green V
    3. Klein RM
    4. Mangun GR

    (2001) Combined expectancies: event-related potentials reveal the early benefits of spatial attention that are obscured by reaction time measures

    Journal of Experimental Psychology: Human Perception and Performance 27:303–317.

    https://doi.org/10.1037/0096-1523.27.2.303

    • PubMed
    • Google Scholar
37. 1. Harvey BM
    2. Dumoulin SO

    (2011) The relationship between cortical magnification factor and population receptive field size in human visual cortex: constancies in cortical architecture

    Journal of Neuroscience 31:13604–13612.

    https://doi.org/10.1523/JNEUROSCI.2572-11.2011

    • PubMed
    • Google Scholar
38. 1. Harvey BM
    2. Dumoulin SO

    (2016) Visual motion transforms visual space representations similarly throughout the human visual hierarchy

    NeuroImage 127:173–185.

    https://doi.org/10.1016/j.neuroimage.2015.11.070

    • PubMed
    • Google Scholar
39. 1. Hayden BY
    2. Gallant JL

    (2005) Time course of attention reveals different mechanisms for spatial and feature-based attention in area V4

    Neuron 47:637–643.

    https://doi.org/10.1016/j.neuron.2005.07.020

    • PubMed
    • Google Scholar
40. 1. Hayden BY
    2. Gallant JL

    (2009) Combined effects of spatial and feature-based attention on responses of V4 neurons

    Vision Research 49:1182–1187.

    https://doi.org/10.1016/j.visres.2008.06.011

    • PubMed
    • Google Scholar
41. 1. Hering E

    (1874)

    Grundrisse einer Theorie des Farbensinnes. Sitzungsberichte der osterreichischen Akademie der Wissenschaften, mathematisch-naturwissenschaftliche

    Klasse Abteiling III 70:169–204.

    • Google Scholar
42. 1. Hicks TP
    2. Lee BB
    3. Vidyasagar TR

    (1983) The responses of cells in macaque lateral geniculate nucleus to sinusoidal gratings

    The Journal of Physiology 337:183–200.

    https://doi.org/10.1113/jphysiol.1983.sp014619

    • PubMed
    • Google Scholar
43. 1. Hillyard SA
    2. Münte TF

    (1984) Selective attention to color and location: an analysis with event-related brain potentials

    Perception & Psychophysics 36:185–198.

    https://doi.org/10.3758/BF03202679

    • PubMed
    • Google Scholar
44. 1. Hopf JM
    2. Boelmans K
    3. Schoenfeld MA
    4. Luck SJ
    5. Heinze HJ

    (2004) Attention to features precedes attention to locations in visual search: evidence from electromagnetic brain responses in humans

    Journal of Neuroscience 24:1822–1832.

    https://doi.org/10.1523/JNEUROSCI.3564-03.2004

    • PubMed
    • Google Scholar
45. 1. Hummer A
    2. Ritter M
    3. Tik M
    4. Ledolter AA
    5. Woletz M
    6. Holder GE
    7. Dumoulin SO
    8. Schmidt-Erfurth U
    9. Windischberger C

    (2016) Eyetracker-based gaze correction for robust mapping of population receptive fields

    NeuroImage 142:211–224.

    https://doi.org/10.1016/j.neuroimage.2016.07.003

    • PubMed
    • Google Scholar
46. 1. Ibos G
    2. Freedman DJ

    (2016) Interaction between spatial and feature attention in posterior parietal cortex

    Neuron 91:931–943.

    https://doi.org/10.1016/j.neuron.2016.07.025

    • PubMed
    • Google Scholar
47. 1. Jehee JF
    2. Brady DK
    3. Tong F

    (2011) Attention improves encoding of task-relevant features in the human visual cortex

    Journal of Neuroscience 31:8210–8219.

    https://doi.org/10.1523/JNEUROSCI.6153-09.2011

    • PubMed
    • Google Scholar
48. 1. Jenkinson M
    2. Beckmann CF
    3. Behrens TE
    4. Woolrich MW
    5. Smith SM

    (2012) FSL

    NeuroImage 62:782–790.

    https://doi.org/10.1016/j.neuroimage.2011.09.015

    • PubMed
    • Google Scholar
49. 1. Jerde TA
    2. Merriam EP
    3. Riggall AC
    4. Hedges JH
    5. Curtis CE

    (2012) Prioritized maps of space in human frontoparietal cortex

    Journal of Neuroscience 32:17382–17390.

    https://doi.org/10.1523/JNEUROSCI.3810-12.2012

    • PubMed
    • Google Scholar
50. 1. Kastner S
    2. Pinsk MA

    (2004) Visual attention as a multilevel selection process

    Cognitive, Affective, & Behavioral Neuroscience 4:483–500.

    https://doi.org/10.3758/CABN.4.4.483

    • PubMed
    • Google Scholar
51. 1. Katzner S
    2. Busse L
    3. Treue S

    (2009) Attention to the color of a moving stimulus modulates motion-signal processing in macaque area mt: evidence for a unified attentional system

    Frontiers in Systems Neuroscience 3:12.

    https://doi.org/10.3389/neuro.06.012.2009

    • PubMed
    • Google Scholar
52. 1. Kay KN
    2. Weiner KS
    3. Grill-Spector K

    (2015) Attention reduces spatial uncertainty in human ventral temporal cortex

    Current Biology 25:595–600.

    https://doi.org/10.1016/j.cub.2014.12.050

    • PubMed
    • Google Scholar
53. 1. Kay KN
    2. Yeatman JD

    (2017) Bottom-up and top-down computations in word- and face-selective cortex

    eLife 6:e22341.

    https://doi.org/10.7554/eLife.22341

    • PubMed
    • Google Scholar
54. 1. Kingstone A

    (1992) Combining Expectancies

    The Quarterly Journal of Experimental Psychology Section A 44:69–104.

    https://doi.org/10.1080/14640749208401284

    • Google Scholar
55. 1. Klein BP
    2. Harvey BM
    3. Dumoulin SO

    (2014) Attraction of position preference by spatial attention throughout human visual cortex

    Neuron 84:227–237.

    https://doi.org/10.1016/j.neuron.2014.08.047

    • PubMed
    • Google Scholar
56. 1. Knapen T
    2. de Gee JW
    3. Brascamp J
    4. Nuiten S
    5. Hoppenbrouwers S
    6. Theeuwes J

    (2016) Cognitive and ocular factors jointly determine pupil responses under equiluminance

    Plos One 11:e0155574.

    https://doi.org/10.1371/journal.pone.0155574

    • PubMed
    • Google Scholar
57. 1. Korn CW
    2. Bach DR

    (2016) A solid frame for the window on cognition: Modeling event-related pupil responses

    Journal of Vision 16:28.

    https://doi.org/10.1167/16.3.28

    • PubMed
    • Google Scholar
58. 1. Kravitz DJ
    2. Behrmann M

    (2011) Space-, object-, and feature-based attention interact to organize visual scenes

    Attention, Perception, & Psychophysics 73:2434–2447.

    https://doi.org/10.3758/s13414-011-0201-z

    • PubMed
    • Google Scholar
59. 1. Kumada T

    (2001) Feature-based control of attention: evidence for two forms of dimension weighting

    Perception & Psychophysics 63:698–708.

    https://doi.org/10.3758/BF03194430

    • PubMed
    • Google Scholar
60. 1. Lee S
    2. Papanikolaou A
    3. Logothetis NK
    4. Smirnakis SM
    5. Keliris GA

    (2013) A new method for estimating population receptive field topography in visual cortex

    NeuroImage 81:144–157.

    https://doi.org/10.1016/j.neuroimage.2013.05.026

    • PubMed
    • Google Scholar
61. 1. Leonard CJ
    2. Balestreri A
    3. Luck SJ

    (2015) Interactions between space-based and feature-based attention

    Journal of Experimental Psychology: Human Perception and Performance 41:11–16.

    https://doi.org/10.1037/xhp0000011

    • PubMed
    • Google Scholar
62. 1. Levin N
    2. Dumoulin SO
    3. Winawer J
    4. Dougherty RF
    5. Wandell BA

    (2010) Cortical maps and white matter tracts following long period of visual deprivation and retinal image restoration

    Neuron 65:21–31.

    https://doi.org/10.1016/j.neuron.2009.12.006

    • PubMed
    • Google Scholar
63. 1. Liu J
    2. Wandell BA

    (2005) Specializations for chromatic and temporal signals in human visual cortex

    Journal of Neuroscience 25:3459–3468.

    https://doi.org/10.1523/JNEUROSCI.4206-04.2005

    • PubMed
    • Google Scholar
64. Software

    1. Love J
    2. Selker R
    3. Marsman M
    4. Jamil T
    5. Dropmann D

    (2015)

    JASP Project, version Jasp (version 0.7)

    Amsterdam, The Netherlands.
65. 1. Luck SJ
    2. Chelazzi L
    3. Hillyard SA
    4. Desimone R

    (1997) Neural mechanisms of spatial selective attention in areas V1, V2, and V4 of macaque visual cortex

    Journal of Neurophysiology 77:24–42.

    https://doi.org/10.1152/jn.1997.77.1.24

    • PubMed
    • Google Scholar
66. 1. Mackey WE
    2. Winawer J
    3. Curtis CE

    (2017) Visual field map clusters in human frontoparietal cortex

    eLife 6:e22974.

    https://doi.org/10.7554/eLife.22974

    • PubMed
    • Google Scholar
67. 1. Maunsell JH
    2. Treue S

    (2006) Feature-based attention in visual cortex

    Trends in Neurosciences 29:317–322.

    https://doi.org/10.1016/j.tins.2006.04.001

    • PubMed
    • Google Scholar
68. 1. McAdams CJ
    2. Maunsell JH

    (2000) Attention to both space and feature modulates neuronal responses in macaque area V4

    Journal of Neurophysiology 83:1751–1755.

    https://doi.org/10.1152/jn.2000.83.3.1751

    • PubMed
    • Google Scholar
69. 1. Miconi T
    2. VanRullen R

    (2016) A feedback model of attention explains the diverse effects of attention on neural firing rates and receptive field structure

    PLOS Computational Biology 12:e1004770.

    https://doi.org/10.1371/journal.pcbi.1004770

    • PubMed
    • Google Scholar
70. 1. Moore T
    2. Armstrong KM

    (2003) Selective gating of visual signals by microstimulation of frontal cortex

    Nature 421:370–373.

    https://doi.org/10.1038/nature01341

    • PubMed
    • Google Scholar
71. 1. Motter BC

    (1994) Neural correlates of attentive selection for color or luminance in extrastriate area V4

    The Journal of Neuroscience 14:2178–2189.

    https://doi.org/10.1523/JNEUROSCI.14-04-02178.1994

    • PubMed
    • Google Scholar
72. 1. Müller MM
    2. Andersen S
    3. Trujillo NJ
    4. Valdés-Sosa P
    5. Malinowski P
    6. Hillyard SA

    (2006) Feature-selective attention enhances color signals in early visual areas of the human brain

    PNAS 103:14250–14254.

    https://doi.org/10.1073/pnas.0606668103

    • PubMed
    • Google Scholar
73. 1. Nordfang M
    2. Staugaard C
    3. Bundesen C

    (2018) Attentional weights in vision as products of spatial and nonspatial components

    Psychonomic Bulletin & Review 25:1043–1051.

    https://doi.org/10.3758/s13423-017-1337-1

    • PubMed
    • Google Scholar
74. 1. O'Craven KM
    2. Downing PE
    3. Kanwisher N

    (1999) fMRI evidence for objects as the units of attentional selection

    Nature 401:584–587.

    https://doi.org/10.1038/44134

    • PubMed
    • Google Scholar
75. 1. Patzwahl DR
    2. Treue S

    (2009) Combining spatial and feature-based attention within the receptive field of MT neurons

    Vision Research 49:1188–1193.

    https://doi.org/10.1016/j.visres.2009.04.003

    • PubMed
    • Google Scholar
76. 1. Pedregosa F
    2. Eickenberg M
    3. Ciuciu P
    4. Thirion B
    5. Gramfort A

    (2015) Data-driven HRF estimation for encoding and decoding models

    NeuroImage 104:209–220.

    https://doi.org/10.1016/j.neuroimage.2014.09.060

    • PubMed
    • Google Scholar
77. 1. Peirce JW

    (2008) Generating stimuli for neuroscience using psychoPy

    Frontiers in Neuroinformatics 2:10.

    https://doi.org/10.3389/neuro.11.010.2008

    • PubMed
    • Google Scholar
78. 1. Pestilli F
    2. Carrasco M

    (2005) Attention enhances contrast sensitivity at cued and impairs it at uncued locations

    Vision Research 45:1867–1875.

    https://doi.org/10.1016/j.visres.2005.01.019

    • PubMed
    • Google Scholar
79. 1. Poldrack RA

    (2017) Precision neuroscience: Dense sampling of individual brains

    Neuron 95:727–729.

    https://doi.org/10.1016/j.neuron.2017.08.002

    • PubMed
    • Google Scholar
80. 1. Posner MI
    2. Snyder CR
    3. Davidson BJ

    (1980) Attention and the detection of signals

    Journal of Experimental Psychology: General 109:160–174.

    https://doi.org/10.1037/0096-3445.109.2.160

    • PubMed
    • Google Scholar
81. 1. Powell MJD

    (1964) An efficient method for finding the minimum of a function of several variables without calculating derivatives

    The Computer Journal 7:155–162.

    https://doi.org/10.1093/comjnl/7.2.155

    • Google Scholar
82. 1. Puckett AM
    2. DeYoe EA

    (2015) The attentional field revealed by single-voxel modeling of fMRI time courses

    Journal of Neuroscience 35:5030–5042.

    https://doi.org/10.1523/JNEUROSCI.3754-14.2015

    • PubMed
    • Google Scholar
83. 1. Reynolds JH
    2. Pasternak T
    3. Desimone R

    (2000) Attention increases sensitivity of V4 neurons

    Neuron 26:703–714.

    https://doi.org/10.1016/S0896-6273(00)81206-4

    • PubMed
    • Google Scholar
84. 1. Reynolds JH
    2. Heeger DJ

    (2009) The normalization model of attention

    Neuron 61:168–185.

    https://doi.org/10.1016/j.neuron.2009.01.002

    • PubMed
    • Google Scholar
85. 1. Rizzolatti G
    2. Riggio L
    3. Dascola I
    4. Umiltá C

    (1987) Reorienting attention across the horizontal and vertical meridians: evidence in favor of a premotor theory of attention

    Neuropsychologia 25:31–40.

    https://doi.org/10.1016/0028-3932(87)90041-8

    • PubMed
    • Google Scholar
86. 1. Rosenholtz R

    (2016) Capabilities and limitations of peripheral vision

    Annual Review of Vision Science 2:437–457.

    https://doi.org/10.1146/annurev-vision-082114-035733

    • PubMed
    • Google Scholar
87. 1. Rossi AF
    2. Paradiso MA

    (1995) Feature-specific effects of selective visual attention

    Vision Research 35:621–634.

    https://doi.org/10.1016/0042-6989(94)00156-G

    • PubMed
    • Google Scholar
88. 1. Saenz M
    2. Buracas GT
    3. Boynton GM

    (2002) Global effects of feature-based attention in human visual cortex

    Nature Neuroscience 5:631–632.

    https://doi.org/10.1038/nn876

    • PubMed
    • Google Scholar
89. 1. Sàenz M
    2. Buraĉas GT
    3. Boynton GM

    (2003) Global feature-based attention for motion and color

    Vision Research 43:629–637.

    https://doi.org/10.1016/S0042-6989(02)00595-3

    • PubMed
    • Google Scholar
90. 1. Savitzky A
    2. Golay MJE

    (1964) Smoothing and differentiation of data by simplified least squares procedures

    Analytical Chemistry 36:1627–1639.

    https://doi.org/10.1021/ac60214a047

    • Google Scholar
91. 1. Schiller PH
    2. Malpeli JG

    (1978) Functional specificity of lateral geniculate nucleus laminae of the rhesus monkey

    Journal of Neurophysiology 41:788–797.

    https://doi.org/10.1152/jn.1978.41.3.788

    • PubMed
    • Google Scholar
92. 1. Schneider KA
    2. Richter MC
    3. Kastner S

    (2004) Retinotopic organization and functional subdivisions of the human lateral geniculate nucleus: a high-resolution functional magnetic resonance imaging study

    Journal of Neuroscience 24:8975–8985.

    https://doi.org/10.1523/JNEUROSCI.2413-04.2004

    • PubMed
    • Google Scholar
93. 1. Schoenfeld MA
    2. Hopf JM
    3. Martinez A
    4. Mai HM
    5. Sattler C
    6. Gasde A
    7. Heinze HJ
    8. Hillyard SA

    (2007) Spatio-temporal analysis of feature-based attention

    Cerebral Cortex 17:2468–2477.

    https://doi.org/10.1093/cercor/bhl154

    • PubMed
    • Google Scholar
94. 1. Schoenfeld MA
    2. Hopf JM
    3. Merkel C
    4. Heinze HJ
    5. Hillyard SA

    (2014) Object-based attention involves the sequential activation of feature-specific cortical modules

    Nature Neuroscience 17:619–624.

    https://doi.org/10.1038/nn.3656

    • PubMed
    • Google Scholar
95. 1. Serences JT
    2. Boynton GM

    (2007) Feature-based attentional modulations in the absence of direct visual stimulation

    Neuron 55:301–312.

    https://doi.org/10.1016/j.neuron.2007.06.015

    • PubMed
    • Google Scholar
96. 1. Sheremata SL
    2. Silver MA

    (2015) Hemisphere-dependent attentional modulation of human parietal visual field representations

    Journal of Neuroscience 35:508–517.

    https://doi.org/10.1523/JNEUROSCI.2378-14.2015

    • PubMed
    • Google Scholar
97. 1. Shulman GL
    2. d'Avossa G
    3. Tansy AP
    4. Corbetta M

    (2002) Two attentional processes in the parietal lobe

    Cerebral Cortex 12:1124–1131.

    https://doi.org/10.1093/cercor/12.11.1124

    • PubMed
    • Google Scholar
98. 1. Silva MF
    2. Brascamp JW
    3. Ferreira S
    4. Castelo-Branco M
    5. Dumoulin SO
    6. Harvey BM

    (2018) Radial asymmetries in population receptive field size and cortical magnification factor in early visual cortex

    NeuroImage 167:41–52.

    https://doi.org/10.1016/j.neuroimage.2017.11.021

    • PubMed
    • Google Scholar
99. 1. Silver MA
    2. Ress D
    3. Heeger DJ

    (2005) Topographic maps of visual spatial attention in human parietal cortex

    Journal of Neurophysiology 94:1358–1371.

    https://doi.org/10.1152/jn.01316.2004

    • PubMed
    • Google Scholar
100. 1. Snyder LH
     2. Grieve KL
     3. Brotchie P
     4. Andersen RA

     (1998) Separate body- and world-referenced representations of visual space in parietal cortex

     Nature 394:887–891.

     https://doi.org/10.1038/29777

     • PubMed
     • Google Scholar
101. 1. Sprague TC
     2. Serences JT

     (2013) Attention modulates spatial priority maps in the human occipital, parietal and frontal cortices

     Nature Neuroscience 16:1879–1887.

     https://doi.org/10.1038/nn.3574

     • PubMed
     • Google Scholar
102. 1. Swisher JD
     2. Halko MA
     3. Merabet LB
     4. McMains SA
     5. Somers DC

     (2007) Visual topography of human intraparietal sulcus

     Journal of Neuroscience 27:5326–5337.

     https://doi.org/10.1523/JNEUROSCI.0991-07.2007

     • PubMed
     • Google Scholar
103. 1. Szczepanski SM
     2. Pinsk MA
     3. Douglas MM
     4. Kastner S
     5. Saalmann YB

     (2013) Functional and structural architecture of the human dorsal frontoparietal attention network

     PNAS 110:15806–15811.

     https://doi.org/10.1073/pnas.1313903110

     • PubMed
     • Google Scholar
104. 1. Theeuwes J
     2. Van der Burg E

     (2007) The role of spatial and nonspatial information in visual selection

     Journal of Experimental Psychology: Human Perception and Performance 33:1335–1351.

     https://doi.org/10.1037/0096-1523.33.6.1335

     • PubMed
     • Google Scholar
105. 1. Tootell RB
     2. Hadjikhani N
     3. Hall EK
     4. Marrett S
     5. Vanduffel W
     6. Vaughan JT
     7. Dale AM

     (1998) The retinotopy of visual spatial attention

     Neuron 21:1409–1422.

     https://doi.org/10.1016/S0896-6273(00)80659-5

     • PubMed
     • Google Scholar
106. 1. Treue S
     2. Martínez Trujillo JC
     3. Trujillo JCM

     (1999) Feature-based attention influences motion processing gain in macaque visual cortex

     Nature 399:575–579.

     https://doi.org/10.1038/21176

     • PubMed
     • Google Scholar
107. 1. Treue S
     2. Maunsell JH

     (1996) Attentional modulation of visual motion processing in cortical areas MT and MST

     Nature 382:539–541.

     https://doi.org/10.1038/382539a0

     • PubMed
     • Google Scholar
108. 1. Trotter Y
     2. Celebrini S

     (1999) Gaze direction controls response gain in primary visual-cortex neurons

     Nature 398:239–242.

     https://doi.org/10.1038/18444

     • PubMed
     • Google Scholar
109. Software

     1. van Es DM
     2. Knapen T

     (2018) pRF_attention_analysis

     Github.

     https://github.com/daanvanes/PRF_attention_analysis
110. 1. Van Opstal AJ
     2. Hepp K
     3. Suzuki Y
     4. Henn V

     (1995) Influence of eye position on activity in monkey superior colliculus

     Journal of Neurophysiology 74:1593–1610.

     https://doi.org/10.1152/jn.1995.74.4.1593

     • PubMed
     • Google Scholar
111. 1. Vo VA
     2. Sprague TC
     3. Serences JT

     (2017) Spatial tuning shifts increase the discriminability and fidelity of population codes in visual cortex

     The Journal of Neuroscience : The Official Journal of the Society for Neuroscience 37:3386–3401.

     https://doi.org/10.1523/JNEUROSCI.3484-16.2017

     • PubMed
     • Google Scholar
112. 1. Watson AB
     2. Pelli DG

     (1983) QUEST: a Bayesian adaptive psychometric method

     Perception & Psychophysics 33:113–120.

     https://doi.org/10.3758/BF03202828

     • PubMed
     • Google Scholar
113. 1. Wegener D
     2. Ehn F
     3. Aurich MK
     4. Galashan FO
     5. Kreiter AK

     (2008) Feature-based attention and the suppression of non-relevant object features

     Vision Research 48:2696–2707.

     https://doi.org/10.1016/j.visres.2008.08.021

     • PubMed
     • Google Scholar
114. 1. White AL
     2. Carrasco M

     (2011) Feature-based attention involuntarily and simultaneously improves visual performance across locations

     Journal of Vision 11:15.

     https://doi.org/10.1167/11.6.15

     • PubMed
     • Google Scholar
115. 1. White AL
     2. Rolfs M
     3. Carrasco M

     (2015) Stimulus competition mediates the joint effects of spatial and feature-based attention

     Journal of Vision 15:7.

     https://doi.org/10.1167/15.14.7

     • PubMed
     • Google Scholar
116. 1. Winawer J
     2. Horiguchi H
     3. Sayres RA
     4. Amano K
     5. Wandell BA

     (2010) Mapping hV4 and ventral occipital cortex: the venous eclipse

     Journal of Vision 10:.

     https://doi.org/10.1167/10.5.1

     • PubMed
     • Google Scholar
117. 1. Wolfe JM
     2. Butcher SJ
     3. Lee C
     4. Hyle M

     (2003) Changing your mind: on the contributions of top-down and bottom-up guidance in visual search for feature singletons

     Journal of Experimental Psychology: Human Perception and Performance 29:483–502.

     https://doi.org/10.1037/0096-1523.29.2.483

     • PubMed
     • Google Scholar
118. 1. Wolfe JM
     2. Horowitz TS

     (2004) What attributes guide the deployment of visual attention and how do they do it?

     Nature Reviews Neuroscience 5:495–501.

     https://doi.org/10.1038/nrn1411

     • PubMed
     • Google Scholar
119. 1. Womelsdorf T
     2. Anton-Erxleben K
     3. Pieper F
     4. Treue S

     (2006) Dynamic shifts of visual receptive fields in cortical area MT by spatial attention

     Nature Neuroscience 9:1156–1160.

     https://doi.org/10.1038/nn1748

     • PubMed
     • Google Scholar
120. 1. Womelsdorf T
     2. Anton-Erxleben K
     3. Treue S

     (2008) Receptive field shift and shrinkage in macaque middle temporal area through attentional gain modulation

     Journal of Neuroscience 28:8934–8944.

     https://doi.org/10.1523/JNEUROSCI.4030-07.2008

     • PubMed
     • Google Scholar
121. 1. Yeshurun Y
     2. Carrasco M

     (1998) Attention improves or impairs visual performance by enhancing spatial resolution

     Nature 396:72–75.

     https://doi.org/10.1038/23936

     • PubMed
     • Google Scholar
122. 1. Yeshurun Y
     2. Carrasco M

     (1999)

     Spatial attention improves performance in spatial resolution tasks

     Vision Research 39:293–306.

     • PubMed
     • Google Scholar
123. 1. Yeshurun Y
     2. Carrasco M

     (2000) The locus of attentional effects in texture segmentation

     Nature Neuroscience 3:622–627.

     https://doi.org/10.1038/75804

     • PubMed
     • Google Scholar
124. 1. Yeshurun Y
     2. Montagna B
     3. Carrasco M

     (2008) On the flexibility of sustained attention and its effects on a texture segmentation task

     Vision Research 48:80–95.

     https://doi.org/10.1016/j.visres.2007.10.015

     • PubMed
     • Google Scholar
125. 1. Zanto TP
     2. Rubens MT
     3. Bollinger J
     4. Gazzaley A

     (2010) Top-down modulation of visual feature processing: the role of the inferior frontal junction

     NeuroImage 53:736–745.

     https://doi.org/10.1016/j.neuroimage.2010.06.012

     • PubMed
     • Google Scholar
126. 1. Zhang W
     2. Luck SJ

     (2009) Feature-based attention modulates feedforward visual processing

     Nature Neuroscience 12:24–25.

     https://doi.org/10.1038/nn.2223

     • PubMed
     • Google Scholar
127. 1. Zhou H
     2. Desimone R

     (2011) Feature-based attention in the frontal eye field and area V4 during visual search

     Neuron 70:1205–1217.

     https://doi.org/10.1016/j.neuron.2011.04.032

     • PubMed
     • Google Scholar
128. 1. Zuiderbaan W
     2. Harvey BM
     3. Dumoulin SO

     (2012) Modeling center-surround configurations in population receptive fields using fMRI

     Journal of Vision 12:10.

     https://doi.org/10.1167/12.3.10

     • PubMed
     • Google Scholar

Thank you for submitting your article "Spatial sampling in human visual cortex is modulated by both spatial and feature-based attention" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Sabine Kastner as the Senior Editor. The following individual involved in review of your submission has agreed to reveal his identity: Michael Silver (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This study measured effects of feature-based attention on pRF eccentricity and size changes related to spatial attention. The authors investigated the hypothesis that feature-based attention optimizes the attended feature's spatial sampling by using a pRF mapping stimulus that varied along two feature dimensions: color and temporal frequency, and they found that attention to color showed stronger modulation of pRF eccentricity and size change compared to attention to temporal frequency. The authors interpret these results with an attentional gain field model by noting that receptive fields preferring color are typically smaller and more foveal compared to those preferring temporal frequency, consistent with these receptive fields having greater changes in eccentricity. They further demonstrate that attentional gain fields were smaller for the Attend Color condition, indicating greater precision of the attentional gain field for color receptive fields. The editors were impressed with this elegant and well-written paper that contains thorough and principled analysis procedures. However, they were concerned regarding some of the statistical analyses (see details below). Also, the authors should clarify how the results advance the field's knowledge about effects of spatial and feature-based attention (and their interaction). The application of the attentional gain field model to these findings was perceived as a strength. Overall, the editors felt that major revisions will be necessary to strengthen the study, so that it can be further pursued at eLife.

Essential revisions:

1) Feature-based attention has long been thought to have modulatory effects on spatial attention (Saenz, Buračas and Boynton, 2002; Maunsell and Treue, 2006). The current study supports this idea by reporting specific analyses where color and temporal frequency are used as attended features and spatial resampling is assessed by measuring populations receptive fields. Despite the effort, the findings of the study are mostly confirmatory in nature. The manuscript would benefit from a reorganized presentation of the results to highlight the novel contributions of the study as opposed to confirmation of prior findings.

2) There isn't adequate information about how the target features were selected. Why should we be interested in attending to temporal frequency versus color in the context of this study? (It is also unclear/unexplained why blue-yellow and cyan-magenta were selected as representative colors.) Would anything change if different colors, a broader range of temporal frequencies, or entirely different feature sets (e.g., orientation) were used?

3) It is reported that pRF changes for Attend Color case are larger compared to Attend Temporal Frequency case. It would, however, be better if the difference could be presented in a more conspicuous way. For instance, the way it is presented in Figure 5A gives the impression that the difference is almost negligible. If there is merely a slight difference, more detailed discussion is warranted regarding the reasons underlying pRF changes when attending to color vs. temporal frequency.

4) Related to the previous comment: When the regions primarily engaged by the two distinct features compared lie at different points in the hierarchy of visual cortex, the strength of retinotopic representations (and thus spatial) will inevitably vary. Wouldn't that introduce a bias in assessment of interactions between spatial and feature-based attention?

5) There is too much variability across subjects in the attentional gain field sizes of temporal frequency vs. color (Figure 7). Given that there appears to be poor consistency across subjects for this measure, this figure does not convey information in a convincing way.

6) The authors combine pRF eccentricity and size to obtain a single index, because they claim there's a high correlation between the two. We don't think this is warranted. For instance, Table 22 shows that there's a significant correlation between eccentricity and size changes in V1 for only 2 out of 5 subjects, and the average correlation is not very high for this ROI either (R=0.461). This also makes it questionable to present the results using voxels from a "combined ROI", given the large degree of variability across individual ROIs.

7) In most of the supplementary figures (e.g., the first one), results for individual subjects are not shown at the same scale across subjects, which makes them confusing. Also, there's barely any change in the pRF size with increasing eccentricity for some ROIs of some subjects (e.g., V1 of s5). Some extended treatment and discussion of the consistency of the presented results across subjects would greatly benefit the manuscript.

8) We found it difficult that when statistical analyses are performed across voxels bootstrap tests were applied, and yet across subject tests were based on ANOVA or t-tests. If the same type of metric is evaluated in both cases, shouldn't you make the same assumptions about the distribution of the metric?

9) More detailed description of the outlier rejection method is needed, as this changes the set of voxels over which the results are reported. This procedure can potentially introduce biases within and across analyses, so it must be motivated very clearly. It is uncommon to reject voxels like this. If the goal is to report robust estimates of central tendency and variance of parameter estimates, then why not calculate them based on median and IQR?

10) It is stated that explained variance during pRF estimation is used to weigh the contributions of individual voxels to estimated parameters. This is understandable, but we would have expected to see actual reports of the explained variance across ROIs as well. This could help readers to judge and interpret the differences observed across ROIs.

11) The stimulus aperture size comprises a very small proportion of the visual field and is also much smaller than those used in other fMRI studies of attentional modulation of pRFs. The authors should provide more discussion about this issue, particularly with respect to the generalizability of their findings. Also, it seems problematic to make the distinction between parafoveal and peripheral pRFs, given that the locations of these pRFs only differ by a couple degrees of visual angle.

12) Figure 8 shows that behavioral accuracy (proportion correct trials) was approximately 80% for the different tasks and eccentricity bins. However, a Quest procedure was used to adaptively change stimulus parameters to generate target performance levels of 83%. Thus, the finding that percent correct did not vary across tasks and eccentricity thresholds is a trivial consequence of the adaptive psychophysical procedure, and statistical analysis providing evidence for the null hypothesis is relatively meaningless. While it would be difficult to directly compare task difficulty across the three tasks, the authors could compute psychophysical thresholds for each task and test whether these are influenced by eccentricity.

Essential revisions:

1) Feature-based attention has long been thought to have modulatory effects on spatial attention (Saenz, Buračas and Boynton, 2002; Maunsell and Treue, 2006). The current study supports this idea by reporting specific analyses where color and temporal frequency are used as attended features and spatial resampling is assessed by measuring populations receptive fields. Despite the effort, the findings of the study are mostly confirmatory in nature. The manuscript would benefit from a reorganized presentation of the results to highlight the novel contributions of the study as opposed to confirmation of prior findings.

We would like to thank the reviewers for pointing out the necessity for better separating our confirmatory from our novel contributions. This comment has prompted us to rewrite the Introduction, Results and Discussion sections of the manuscript. The interaction between spatial and feature-based attention has indeed received generous scientific attention. Nevertheless, it remains elusive whether feature-based attention can influence spatial resampling that results from spatial attention. Although this relationship was predicted by behavioral work (Yeshurun and Carrasco, 1998, 2000; Yeshurun et al., 2008; Barbot and Carrasco, 2017), it has remained untested in electrophysiological recordings or functional neuroimaging. We are the first to directly assess this interaction in the present manuscript. In order to appreciate this novel contribution, we argue that it is of great importance to first present readers with a detailed description of the spatial resampling that resulted from spatial attention (which is mainly of confirmatory nature). Then, we can address how this is modulated by feature based attention. For this reason, we did not consider it opportune to change the order of presentation within the Results section. Instead, we implemented various changes to the text in order to better manage reader's expectations. For this, we first more concretely stress the novelty of our study in the Introduction:

“The studies mentioned above investigated modulatory effects of feature-based attention on spatial attention by measuring changes in response amplitude (e.g. ERP/firing rate). […] In the current study, we put this hypothesis to the test by measuring the brain's representation of space under conditions of differential attention.”

In addition, we added a paragraph in the Results section to reiterate this principal and novel aim, and to manage the reader's expectations about confirmatory vs. novel results:

“Our principal aim is to understand whether feature-based attention influences spatial resampling induced by spatial attention. […] Finally, we relate these feature-based attentional modulations to (1) bottom-up feature preference and (2) to differences in the spatial scale at which color and temporal frequency are processed.”

Finally, we also changed the order of the Discussion such that we now first discuss our novel findings, before we move on to relating our more confirmatory findings to the literature.

2) There isn't adequate information about how the target features were selected. Why should we be interested in attending to temporal frequency versus color in the context of this study? (It is also unclear/unexplained why blue-yellow and cyan-magenta were selected as representative colors.) Would anything change if different colors, a broader range of temporal frequencies, or entirely different feature sets (e.g., orientation) were used?

We will first clarify our reasoning for opting for color and temporal frequency. Next, we will argue why our results should not be specific for the specific features used in our experiment.

There are two important reasons for why we studied attending color versus temporal frequency. First, behavioral work suggested that the spatial scale at which features are processed should determine the degree of spatial resampling (Yeshurun and Carrasco, 1998, 2000; Yeshurun et al., 2008; Barbot and Carrasco, 2017). We therefore opted to use features that are known to be processed at different spatial scales. We previously only stressed a single factor that contributes such differential spatial sampling between these features: the eccentricity dependence of both pRF size and relative preference for color compared to temporal frequency. However, the cortical areas that preferentially process color and temporal frequency (i.e. MT+ compared to hV4; Liu and Wandell, 2005; Brouwer and Heeger, 2009, 2013; Winawer et al., 2010) differ in their average receptive field sizes (Amano et al., 2009; Winawer et al., 2010). Therefore, temporal frequency compared to color is processed on average by neurons with larger receptive fields. Finally, a third factor contributing to differential spatial sampling is that color compared to temporal frequency depends more on the parvocellular versus the magnocellular pathway, where spatial pooling is more fine-grained (Schiller and Malpeli, 1978; Hicks et al., 1983; Denison et al., 2014). In addition to this differential spatial sampling argument, the second main reason for opting for color and temporal frequency is that they are known to be processed in a distributed manner throughout the cortex (color: hV4 / temporal frequency: MT+). This provides an anatomically segregated handle for investigating the relationship between bottom-up feature preference and feature-based attentional modulations. As previously noted in our Materials and methods section, we specifically chose temporal frequency and not coherent motion, as coherent motion signals have been shown to influence pRF measurements (Harvey and Dumoulin, 2016). Also, it was previously shown that attention can be directed to these features (Wolfe and Horowitz, 2004; Cass et al., 2011). We now added a paragraph to our Introduction detailing this multi-faceted argument for why attending color and temporal frequency are of particular interest to our study:

“One important reason for studying the effects of attention to color and temporal frequency is that they are known to be processed at different spatial scales. […] Moreover, it was previously shown that attention can be directed to both feature domains (Wolfe and Horowitz, 2004; Cass et al., 2011).”

We also incorporated the two additional factors that contribute to differential spatial resampling between color and temporal frequency (i.e. parvocellular vs. magnocellular and offset in pRF size between hV4/MT+) in our Results section:

“What could then explain the fact that attending color in the stimulus induced greater changes in spatial sampling? […] In sum, the greater amount of spatial resampling when attending color can be parsimoniously explained by color being sampled by relatively smaller pRFs.”

And we discuss these two factors in our Discussion:

“Previous behavioral reports suggest that the spatial scale at which an attended feature is processed should influence the degree of spatial resampling (Yeshurun and Carrasco, 1998, 2000; Yeshurun et al., 2008; Barbot and Carrasco, 2017). […] Finally, both the current and previous studies show that pRF size (Dumoulin and Wandell, 2008) and color compared to temporal frequency preference (Curcio et al., 1990; Azzopardi et al., 1999; Brewer et al., 2005) vary across eccentricity such that foveal voxels have smaller pRFs and are more color sensitive.”

Previously, as mentioned above, we mainly emphasized the eccentricity gradient of both feature-preference and pRF size as underlying the finer spatial scale of color compared to temporal frequency processing. The initial version of our manuscript therefore contained a note about alternative eccentricity-dependent effects of attention. However, as we now stress two other factors that contribute to differences in spatial sampling between the two features, we currently argue that including this note leads to more distraction than that it is clarifying. We therefore removed these sentences from the Discussion:

“It must however be noted that as pRF size is closely related to pRF eccentricity, it is conceivable that some additional eccentricity dependent attentional modulation influenced our results. […] Therefore, although differential influences of attention across eccentricities have been observed before in the brain (e.g. Roberts et al., 2007; Bressler et al., 2013), these are likely brought about by differences in pRF size.”

Regarding the choice of specific colors, we would first like to note that after additional consideration we deem the name 'green' as more appropriate for the color previously indicated with 'cyan' (RGB color: (0,255,128)). We have therefore renamed the color pairs: blue/yellow and green/magenta. In our experiment, we opted for color pairs that were opposite in terms of color opponency (blue/yellow and green/magenta, Hering (1874)). This ensured that the Gabor peaks and troughs were made up of colors that provided strong color contrast. In addition, this meant that we presented stimuli to all color channels, in effect serving as a potent and full range color stimulus. We added these lines to the Materials and methods section:

“We opted for color pairs that were opposite in terms of color opponency (Hering (1874)). […] In addition, this meant that we presented stimuli to all color channels, in effect serving as a potent and full range color stimulus.”

Similarly, we used multiple temporal frequencies (3-7 Hz), in order to serve as a relatively general and broad temporal frequency stimulus. We argued above that it is the difference in spatial scale at which features are processed that influences the degree of spatial resampling. This implies that the particular choice of colors or temporal frequencies can only influence the degree of resampling if one has a particular reason for believing that different feature values are processed at different spatial scales. Attending these particular feature values in different attentional conditions should then lead to differential spatial resampling. We would like to point out however that in our experiment, attention was always devoted to all colors simultaneously within the Attend Color condition and to all temporal frequencies in the Attend TF condition. We therefore argue that our results reflect a general situation of attending color compared to temporal frequency. Finally, our results should indeed generalize to attending other feature dimensions (e.g. attending faces versus letters), as long as these features differ in the spatial scale of the visual system’s sensitivities to them. We added the abovementioned logic to the Discussion:

“In sum, we suggest that the greater degree of spatial resampling when attending color compared to temporal frequency can be explained by the difference in spatial scale at which these features are processed. […] This includes attending different feature values such as high compared to low spatial frequency, or attending different feature dimensions such as faces (broader spatial scale) versus letters (finer spatial scale).”

3) It is reported that pRF changes for Attend Color case are larger compared to Attend Temporal Frequency case. It would, however, be better if the difference could be presented in a more conspicuous way. For instance, the way it is presented in Figure 5A gives the impression that the difference is almost negligible. If there is merely a slight difference, more detailed discussion is warranted regarding the reasons underlying pRF changes when attending to color vs. temporal frequency.

Indeed, the absolute difference as presented in Figure 6A is relatively small. It should be noted however that this measure averages over the many changes in pRF parameters across attention to all possible bar positions. The magnitude of this measure should therefore not be interpreted as absolute shifts in degrees towards single attentional locations. A more interpretable measure is that of the feature based attentional modulation index (FAMI). This measure takes into account the average shift over all bar locations and computes feature based attentional modulations relative to this baseline. This FAMI measure is a contrast-like measure (difference divided by sum), and yields values around 0.05. This means that pRF changes were on average 10% larger when attending color compared to when attending temporal frequency. In fact, in some ROIs, the FAMI reaches values of 0.20, which roughly corresponds to 50% stronger pRF changes when attending color compared to temporal frequency. We would argue that an increase in pRF changes of 10-50% is not negligible and in fact is likely to have a substantial effect on visual information processing and perception. We added this logic to the Results section:

“Specifically, the feature AMI was around.05 on average across ROIs. As the feature AMI is a contrast measure where difference is divided by the sum, this roughly corresponds to 10% stronger pRF changes when attending color compared to temporal frequency. In some ROIs (V3AB/IPS0, see Figure 6—figure supplement 2), the FAMI reached values of.20, which roughly corresponds to 50% stronger pRF changes when attending color compared to temporal frequency.”

4) Related to the previous comment: When the regions primarily engaged by the two distinct features compared lie at different points in the hierarchy of visual cortex, the strength of retinotopic representations (and thus spatial) will inevitably vary. Wouldn't that introduce a bias in assessment of interactions between spatial and feature-based attention?

We agree that we did not sufficiently address the differences in visual hierarchy at which the two attended features are processed. In the current version of the manuscript, we now explicitly include this argument as one of the factors contributing to differences in the spatial scale at which the two attended features are processed (see our reply to Essential revision point 2). Indeed, it should be expected that offsets in pRF size should lead to differential pRF changes. Specifically, Gaussian interaction models of attention (Klein et al., 2014) suggest that the larger the stimulus drive (i.e. the pRF outside influence of attention), the greater the impact of attention. In correspondence, we observed that absolute pRF shifts were larger in areas with larger average pRFs (see Figures 3 and 4). Importantly however, our assessment of feature based attentional modulation is a contrast measure (AMI), relative to the absolute pRF modulation resulting from differential spatial attention. This analysis therefore controls for any offset in pRF size. Moreover, any difference in pRF size is also taken into account by the attentional gain field modeling (i.e. it is input to the model). Together, this means that differences in pRF size between visual areas that preferentially process the attended feature should not lead to biases in the assessment of interactions between spatial and feature based attention in our analyses. Nevertheless, (as we argue above) these offsets of pRF size between these regions does contribute to explaining why differential spatial resampling is required when attending color compared to temporal frequency.

5) There is too much variability across subjects in the attentional gain field sizes of temporal frequency vs. color (Figure 7). Given that there appears to be poor consistency across subjects for this measure, this figure does not convey information in a convincing way.

This comment has prompted us to change the presentation of our data in this figure for improved clarity (current Figure 8). Specifically, both previous findings (Klein et al., 2014) and our data (Figure 5) show that a single attentional gain field operates throughout visual cortex. In addition, we did not find that feature-based attentional modulation was related to feature preference across visual areas (Figure 6E). We therefore summarized feature-based attentional modulations of the attentional gain field across visual areas as the median across individually fitted ROIs (previously dark shaded area) and as the combined ROI (i.e. fitted on data from all ROIs together). We clarified this in our manuscript:

“Both our data (Figure 5) and previous findings (Klein et al., 2014) showed that a single attentional gain field affects the different visual regions similarly. […] We therefore analyzed feature-based attentional modulations of the attentional gain field both on data from all ROIs fitted together (the 'combined ROI'), and as the median across individually fitted ROIs.”

6) The authors combine pRF eccentricity and size to obtain a single index, because they claim there's a high correlation between the two. We don't think this is warranted. For instance, Table 22 shows that there's a significant correlation between eccentricity and size changes in V1 for only 2 out of 5 subjects, and the average correlation is not very high for this ROI either (R=0.461). This also makes it questionable to present the results using voxels from a "combined ROI", given the large degree of variability across individual ROIs.

In order to verify whether our feature-based attentional modulation (feature AMI) results were not specific to our combined measure of eccentricity and size changes, we repeated this analysis using pRF size and eccentricity separately and report them in the manuscript (see the novel Figure 6—figure supplement 3). This analysis returned highly similar results compared to our initial combined pRF size and eccentricity feature-based AMI. In our opinion, it is warranted to combine the pRF size and eccentricity differences into a single measure of attentional modulation. We added a reference to this figure in the Results section:

“Computing the AMI with either pRF eccentricity or size changes separately (i.e. not as a combined measure) returned similar results (see Figure 6—figure supplement 3).”

We agree that it is striking that the correlation between eccentricity and size is somewhat lower in V1 compared to the other ROIs. We suspect that this might be due to the fact that pRFs are very small in V1. This in turn means that absolute attention-induced changes in pRF eccentricity and size are also very small and therefore harder to measure with high precision. In addition, we showed above that feature-based attentional pRF modulations are present for both eccentricity and size changes separately. So, although these correlations are smaller in V1, we conclude it is likely that this is not the result of fundamental differences in the effects of attention. We continue to feel that it is meaningful to combine all voxels into a single 'combined' ROI, as it serves to show the predominant effects present across all voxels in the visual brain. We want to stress that the combined ROI is not meant to present results averaged across ROI. Indeed, the different ROIs contribute meaningfully differently to this combined ROI (also see Essential revision point 10). The individual ROI results then serve to specify this overall effect per ROI. In our view, the two analyses serve complementary purposes. For these reasons, we believe that including the combined ROI helps the reader to better appreciate the predominant patterns in the data.

7) In most of the supplementary figures (e.g., the first one), results for individual subjects are not shown at the same scale across subjects, which makes them confusing. Also, there's barely any change in the pRF size with increasing eccentricity for some ROIs of some subjects (e.g., V1 of s5). Some extended treatment and discussion of the consistency of the presented results across subjects would greatly benefit the manuscript.

In order to provide additional assessment of the stability of the results across subjects, we matched the scale of all the individual subject figures. In addition, we performed additional analyses (see novel Figure 2—figure supplement 1, and Tables 1 and 11) on the slope between eccentricity and size in all ROIs and in all individual subjects. Although this relationship was numerically positive in all ROIs in all subjects, it did not cross the significance threshold of α=.05 in some ROIs (VO/V3AB/IPS0/MT+; p’s of. 146,. 073,. 078 and. 058 respectively) in no more than 1 out of 5 subjects. We therefore removed the incorrect 'in all participants' statement from the main text and apologize for its initial inclusion. In order to provide more extended discussion of the results across subjects, we added a novel paragraph to the Results section:

“In order to evaluate the stability of our results we repeated all analyses for individual subjects. […] In addition, these analyses showed that spatial resampling was consistently modulated by feature-based attention across subjects (i.e. feature AMI was on average 0.059 greater than 0 (F_(1,4) = 18.868, p =.012, η2p=.394), and was not different between ROIs (F_(8,32) = 0.631, p =.746, η2p=.066)).”

8) We found it difficult that when statistical analyses are performed across voxels bootstrap tests were applied, and yet across subject tests were based on ANOVA or t-tests. If the same type of metric is evaluated in both cases, shouldn't you make the same assumptions about the distribution of the metric?

Our rationale for using bootstrapping across voxels is that this allowed for the computation of confidence intervals for metrics that can only be derived across a set of voxels (e.g. correlation values). This also allowed us to visualize the distribution of the metric in a data-driven fashion. Letting go of the assumption of normality in bootstrapping thereby increased the specificity of the derived results. We would preferably also have used bootstrapping when calculating statistics over subjects, but bootstrapping is not stable for N<20. We made the additional assumption of normality for across subjects in order to increase stability and robustness of the derived p-values. We thus utilized both bootstrapping and more classical statistics to optimize the balance between sensitivity and robustness. We added these sentences to the Materials and methods section in order to clarify this issue:

“Our rationale for using bootstrapping across voxels and more classical tests across subjects is that while bootstrapping closely represents the actual distribution of the data, it is not stable for N < 20. We therefore made the additional assumption of normality when computing statistics across subjects.”

In order to complete the consistency of using bootstrapping across participants and classical statistics across voxels, we now changed the visualization of confidence intervals in Figure 9A from using bootstrapping to classical statistics. Note that this is merely a matter of visualization, as the Bayesian analyses reported are not affected. Updated CIs are slightly wider, as classical statistics is more conservative than bootstrapping for small Ns.

9) More detailed description of the outlier rejection method is needed, as this changes the set of voxels over which the results are reported. This procedure can potentially introduce biases within and across analyses, so it must be motivated very clearly. It is uncommon to reject voxels like this. If the goal is to report robust estimates of central tendency and variance of parameter estimates, then why not calculate them based on median and IQR?

In computing the AMI, division by values close to zero sometimes resulted in extreme values. In order to remove these values from further analysis, we opted to use a very conservative outlier rejection method, using 5 median absolute deviations as a threshold. We specifically opted to apply the same outlier rejection method to all other analyses for consistency. In order to verify whether outlier rejection had any significant effect on the other analyses, we also repeated all analyses without outlier rejection and found no meaningful changes. Using the median and IQR, as the reviewers suggest, would have made for a suitable alternative. This should have yielded very similar results, as our method similarly rejected outliers based on deviations from the median. Indeed, our approach changed the set of voxels over which the results are reported. Yet, the IQR method would similarly base its results on different subpopulations of voxels per analysis (namely those voxels not too far from the median), without explicitly stating so in changes in N. We therefore argue that our approach is both conservative, consistent (across analyses) and transparent (stating the N). We've added a more detailed description of this outlier rejection method in the Materials and methods section:

“The computation of the attentional modulation index (see below) included a division operation. […] Importantly, in all of these remaining analyses, outlier rejection did not meaningfully alter any result.”

In order to further increase the consistency of our outlier rejection method, we simplified the determination of outliers in the feature-AMI analysis. Previously, outliers were determined based on multiple measures, including the color AMI, the TF AMI, the combined AMI and on feature-preference. Currently, we only determine outliers based on the feature-AMI in this analysis. This increased the number of included voxels and slightly changed the exact statistical values in the feature-AMI analyses (Tables 9 and 25 and legend of Figure 6—figure supplement 4). In maximally 1/5 subjects, this resulted in changes of statistical significance (in V3, LO, and MT+). Specifically, in V3 this resulted in one additional subject showing significantly positive AMI; in MT+ it resulted in 1 fewer subject showing significantly positive AMI; in LO, it resulted in 1 fewer subject showing significantly negative AMI. As a result, none of the subjects show a significantly negative AMI in any ROI. In addition, it did not change any of the conclusions about statistical significance for the ‘super subject’ method nor when computing t- and F-tests over subject values (‘over subjects’ method).

10) It is stated that explained variance during pRF estimation is used to weigh the contributions of individual voxels to estimated parameters. This is understandable, but we would have expected to see actual reports of the explained variance across ROIs as well. This could help readers to judge and interpret the differences observed across ROIs.

We agree. In order to present additional information regarding the distribution of explained variance across the different ROIs, we added two panels to Figure 2 (and added Figure 2—figure supplement 2). The first panel (C) depicts visualizations of the distributions of explained variance across voxels for each ROI. The second panel (D) displays visualizations of the number of voxels per ROI. Combining the information from these two figures provides the reader with a comprehensive overview of the total amount of signal that was present in each ROI. In addition, we added a supplementary figure that details this information per subject, and also shows how many voxels remain in each ROI after the application of the different rejection criteria.

11) The stimulus aperture size comprises a very small proportion of the visual field and is also much smaller than those used in other fMRI studies of attentional modulation of pRFs. The authors should provide more discussion about this issue, particularly with respect to the generalizability of their findings. Also, it seems problematic to make the distinction between parafoveal and peripheral pRFs, given that the locations of these pRFs only differ by a couple degrees of visual angle.

We understand that our stimulus could be considered small compared to the number of degrees of visual angle that spans our visual world (about 5% assuming full visual field of 144 degrees). Even though our stimulus only spanned roughly 5% of the visual field, this corresponds to approximately 25% of V1 surface due to cortical magnification (Adams and Horton, 2003). As this bias for over representing the central part of the visual field increases up the visual hierarchy (Harvey and Dumoulin, 2011), this percentage represents a minimum bound of cortical surface area stimulated by our stimulus. Comparing our stimulus size to other studies of attention and pRFs, we find a maximum stimulus eccentricity of 4.5 degrees in Vo et al., 2017, (corresponding to roughly 30% of V1 surface as approximated based on the results by Adams and Horton, 2003), of 5 degrees eccentricity in Klein et al., 2014, (~33% V1 surface), of 7.5 degrees in Kay et al., 2015, (~45% of V1 surface), and of 14 degrees in Sheremata and Silver, 2015, (~60% of V1 surface). Although our stimulus size is thus relatively small, it is generally comparable to two of these studies in terms of proportion of stimulated V1 (25% in our study compared to 30% and 33% in Vo et al., 2017, and Klein et al., 2014, respectively). Yet, one might argue that 25% is still a low number, and that the stimulus preferably should have been larger. However, increasing stimulus size also increases bar stimulus step size (keeping stimulation time equal). This means that the precision at which pRF properties like position and size can be determined decreases with increasing stimulus size. As we sought to measure changes in spatial sampling with the highest possible fidelity, we opted to keep the stimulus size relatively small in order to maximize our sensitivity for measuring changes in pRF parameters. Finally, we understand that the term 'parafoveal' and 'peripheral' are not correct in absolute terms. We meant to use those terms as relative, rather than absolute measures of eccentricity. We now specifically added a section to clarify this in our Materials and methods section:

“Although our stimulus spanned roughly 5% of the visual field (assuming a visual field of 144 dva), this corresponds to roughly 25% of V1 surface area due to cortical magnification (Adams and Horton, 2003). […] Throughout the Results section, we use the terms 'parafoveal' and 'peripheral' to indicate positions relative to the stimulated portion of visual field.”

12) Figure 8 shows that behavioral accuracy (proportion correct trials) was approximately 80% for the different tasks and eccentricity bins. However, a Quest procedure was used to adaptively change stimulus parameters to generate target performance levels of 83%. Thus, the finding that percent correct did not vary across tasks and eccentricity thresholds is a trivial consequence of the adaptive psychophysical procedure, and statistical analysis providing evidence for the null hypothesis is relatively meaningless. While it would be difficult to directly compare task difficulty across the three tasks, the authors could compute psychophysical thresholds for each task and test whether these are influenced by eccentricity.

We agree with the reviewers that not finding any difference in task performance is to be expected using a Quest staircase procedure. Nevertheless, using this procedure does not necessarily guarantee success. It is for instance possible that the procedure settles at a wrong difficulty level, making this verification necessary. We would argue that Quest and similar procedures are too often used as a panacea without this necessary verification. The analysis presented in the manuscript therefore serve to verify the validity of the Quest procedure in achieving the desired level of accuracy. We changed the introductory text to this part of the Results section to more clearly state this aim:

“Although we used a Quest procedure to equate difficulty across attention conditions and across different levels of eccentricity, it is possible that this procedure stabilized at a faulty difficulty level. In order to verify whether the Quest procedure successfully equated performance we used a similar Bayesian approach, testing whether a model including attention condition (3 levels) and stimulus eccentricity (3 levels) influenced behavioral performance (Figure 9A).”

We would like to note that stimuli were the same across attention conditions, and that our procedure successfully equated task difficulty. These were the chief aims of our behavioral paradigm. Yet, this reviewer comment has prompted us to also analyze the ratio of Gabor elements of either feature value used by the Quest procedure in order to equate difficulty. We placed these analyses in a novel figure supplement (see Figure 9—figure supplement 1). The legend of this supplement reads as the following:

“Figure 9—figure supplement 1.Ratio of Gabor elements of either feature value used by the Quest procedure in order to equate difficulty. […] This could suggest that the greater degree of spatial resampling we observed when attending color discounted the lower relative sensitivity for color in the periphery.”

Author details

1. Daniel Marten van Es

   Behavioural and Movement Sciences, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing

   For correspondence

   daan.van.es@gmail.com

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-7067-6394

2. Jan Theeuwes

   Behavioural and Movement Sciences, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands

   Contribution

   Resources, Funding acquisition, Writing—original draft

   Competing interests

   No competing interests declared

3. Tomas Knapen

   1. Behavioural and Movement Sciences, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands
   2. Spinoza Centre for Neuroimaging, Royal Academy of Sciences, Amsterdam, The Netherlands

   Contribution

   Conceptualization, Resources, Software, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing—original draft, Project administration, Writing—review and editing

   For correspondence

   tknapen@gmail.com

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-5863-8689

• 1,547

  Page views

• 216

  Downloads

• 14

  Citations

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