Recent behavioral and electrophysiological evidence suggests that despite our seamless visual experience, incoming visual information is periodically enhanced for analysis in the visual system (VanRullen, 2016a; VanRullen, 2016b; Zoefel and VanRullen, 2017). This periodic sampling mechanism is proposed to result from the allocation of visual attention (Busch and VanRullen, 2010; Dugué et al., 2016; Dugué and VanRullen, 2017a; VanRullen et al., 2007; Zoefel and VanRullen, 2017), wherein alternating windows of high- and low-attentional resources operate to parcel incoming visual information, similar to the sequential frames that capture film within a video camera (Chakravarthi and Vanrullen, 2012; Vanrullen and Dubois, 2011). Whether stimuli are presented at the appropriate phase (Busch et al., 2009; Mathewson et al., 2009; VanRullen et al., 2007) or location (Dugué et al., 2015; Dugué et al., 2016; Dugué and Vanrullen, 2014; Dugué and VanRullen, 2017a; Dugué et al., 2017b; Huang et al., 2015; Landau and Fries, 2012; Song et al., 2014) of this sampling mechanism has been shown to modulate the accurate detection of a visual stimulus, in stark contrast to our experience of an uninterrupted visual environment.

To date, primary neural evidence for the rhythmic gating of visual processing stems from the dependence of target detection on the pre-target phase of neural oscillations at approximately 7–8 Hz (Busch et al., 2009; Busch and VanRullen, 2010). These spontaneous fluctuations in detection may result from the allocation of visual attention toward a single location (Busch and VanRullen, 2010; Dugué et al., 2015; Spaak et al., 2014; VanRullen, 2016b; Zoefel and VanRullen, 2017), and support the assumption that neural excitability cycles gate and filter incoming information for further processing (Schroeder and Lakatos, 2009; VanRullen, 2013; Zoefel and VanRullen, 2017).

This periodic gating of visual perception is also prominent behaviorally in the time-course of detection accuracy. Spectral analyses applied to high temporal resolution behavioral measures reveal 7–8 Hz modulations in performance following cues to reorient attention (Fiebelkorn et al., 2013), which slow proportionately when attention is divided between two or more locations (e.g. Chen et al., 2017; Holcombe and Chen, 2013; Huang et al., 2015; Landau and Fries, 2012; Landau et al., 2015; VanRullen, 2013). For example, Landau and Fries, 2012 observed that following a cue to reorient attention to either the left or right visual hemifield, target detection oscillated at a 4 Hz counterphase rhythm depending on whether cues were congruent or incongruent with the target location. Critically, this counterphase sampling of visual information persisted at ~4 Hz when attention was directed to two locations on a single object (Fiebelkorn et al., 2013), and when cues to reorient attention were incongruent with target location – requiring a subsequent shift in the allocation of attention to a second location (Huang et al., 2015). These successive fluctuations in target detection and counterphase sampling between locations have led to the suggestion that an intrinsic ~7–8 Hz attentional rhythm can be allocated over space and time in a sequential manner (Dugué et al., 2016; Dugué and VanRullen, 2017a; Fiebelkorn et al., 2013; Holcombe and Chen, 2013; Landau and Fries, 2012; VanRullen, 2013; Zoefel and VanRullen, 2017).

Here, we tested if rhythmic attentional sampling is at play during binocular rivalry. During binocular rivalry, incompatible images are presented to each eye which results in stochastic perceptual alternations, with one image visible at a time while the other is suppressed (Alais, 2012; Alais and Blake, 2005; Maier et al., 2012). In an experiment designed to induce or delay these transitions using auditory and tactile cues, we found that changes in consciousness were occurring rhythmically after the reorientation of attention. These fluctuations occurred depending on whether the crossmodal cue directed attention toward either the dominant or suppressed visual image, resulting in ~8 Hz and ~3.5 Hz oscillations, respectively. Critically, these rhythms were observed in both behavior and the electroencephalogram (EEG), and were absent when cues were not task-relevant. This approximate halving of frequency suggests that when non-visual input is inconsistent with the ongoing visual percept, attentional sampling can flexibly orient away from a consciously perceived image, seemingly ‘searching for’ alternative sensory information to resolve the conflict.

Our findings provide novel evidence that attentional sampling exists during binocular rivalry, demonstrated in both behavior and the electroencephalogram (EEG). Behaviorally, we replicated previous evidence that stimulus-driven cues can cause a switch to previously suppressed visual stimuli when mismatched with the current percept (to bring about congruence), as well as increase the maintenance of a dominant visual image if cues matched perception (Figure 2; Dieter et al., 2015; Lunghi and Alais, 2015; Lunghi et al., 2014). Critically, we found distinct attentional sampling frequencies evident in the time-course of first switches during these cues (Figure 3). When crossmodal cues were incongruent in temporal frequency with the dominant visual stimulus, switches in visual consciousness occurred earlier, and within a distinct ~3.5 Hz rhythm. This 3.5 Hz rhythm is consistent with previous reports of divided attentional sampling between two locations (Fiebelkorn et al., 2013; Landau and Fries, 2012; Landau et al., 2015). However, when crossmodal cues were matched in temporal frequency to the dominant visual stimulus, changes in visual consciousness demonstrated an ~8 Hz rhythm, consistent with periodicities in behavioral measures observed when attending to a single visual location (Fiebelkorn et al., 2013), and suggestions that a cortical 7–8 Hz attentional rhythm may gate visual processing (Busch and VanRullen, 2010; Dugué and VanRullen, 2017a; Fries, 2015). In the EEG (Figures 4 and 5), distinct correlates of these divided and focused attentional sampling frequencies emerged over right fronto-temporal and right parieto-occipital sites, respectively, with ITPC strength at these frequencies correlating with the behaviorally reported change in consciousness across subjects.

Traditionally, top-down, voluntary attention has been thought to have limited control over perceptual dynamics during binocular rivalry; attention may alter dominance durations, but cannot halt the process of perceptual reversals entirely (Chong and Blake, 2006; Chong et al., 2005; Chopin and Mamassian, 2010; Dieter et al., 2016b; Dieter et al., 2015; Dieter et al., 2016b; Mitchell et al., 2004; Paffen and Alais, 2011; for bottom-up control, including crossmodal stimulation, see Conrad et al., 2010; Deroy et al., 2014; Guzman-Martinez et al., 2012; Kang and Blake, 2005; Lunghi and Alais, 2013; Lunghi et al., 2010; Lunghi et al., 2014; van Ee et al., 2009). Our results clearly show additional dependence on the top-down deployment of attention, as without explicit instruction to attend to crossmodal signals, no facilitatory crossmodal effects emerged (see also Jack and Hacker, 2014; Talsma et al., 2010; van Ee et al., 2009). This interaction between low-level stimulus features (temporal frequency) and the allocation of attention indicates the facilitative role of both crossmodal stimuli (Deroy et al., 2014; Deroy et al., 2016) and attention for perceptual transitions during binocular rivalry (Dieter et al., 2016a; Dieter et al., 2015; Dieter and Tadin, 2011; Paffen and Alais, 2011; Zhang et al., 2011b). Our results are consistent with previous research, which has shown that exogenous feature-based cues can bias rivalry dynamics (Dieter et al., 2015; van Ee et al., 2009) and extend these reports by revealing an oscillatory basis to these changes in visual perception.

Our demonstration of these oscillatory changes in visual consciousness, which have been evoked by attended crossmodal cues, are also relevant to current computational models of binocular rivalry (e.g. Laing and Chow, 2002; Li et al., 2017; Wilson, 2003). No current model has accounted for the interaction we observe between attentional allocation and crossmodal stimuli, nor attention as an oscillatory process. Most recently, Li et al., 2017 have described a model for binocular rivalry incorporating attention and mutual inhibition. In their model, attentional modulation is dealt to the sensory representation that has the stronger sensory responses, by providing feedback to monocular-excitatory drives that otherwise increase monotonically with stimulus contrast. Building on our findings, future models could incorporate an oscillatory increase in excitatory drive as a result of periodic, rather than sustained attentional modulation (Fiebelkorn et al., 2018; Helfrich et al., 2018; VanRullen, 2018).

Previous behavioral investigations of attentional sampling have relied upon a brief cue to reorient attention, before estimating the time-course of target detection by densely sampling subject responses over closely spaced target-presentation intervals. Our design is unique in that ‘target-detection’ here is operationalized as the first reported change in visual consciousness for a continuously presented stimulus, resolved at 16.7 ms (or 60 Hz) from 500 ms to 2000 ms following cue-onset.

Past researchers have demonstrated approximately 7–8 Hz fluctuations in perceptual performance following the allocation of visual attention (Dugué et al., 2015; Fiebelkorn et al., 2013; VanRullen, 2013; VanRullen et al., 2007; Zoefel and VanRullen, 2017), commensurate with suggestions that cortical oscillations at approximately 7–8 Hz gate the content of visual perception (Busch and VanRullen, 2010; Dugué and VanRullen, 2017a; Hanslmayr et al., 2013). In our binocular rivalry paradigm, we also observed changes in visual consciousness occurring within an 8 Hz rhythm, yet unique to when cues were congruent with the dominant visual percept at cue onset. By contrast, perceptual sampling has previously been observed at ~4 Hz when cues have encouraged dividing attention between two objects or locations (Dugué et al., 2016; Fiebelkorn et al., 2013; Huang et al., 2015; Landau and Fries, 2012; Landau et al., 2015; Song et al., 2014). As such, the ~3.5 Hz rhythm we observed when crossmodal cues mismatched with the conscious visual percept extends the evidence for divided attentional sampling to binocular rivalry.

From our data alone, we cannot infer whether during conventional binocular rivalry, attention samples at 8 or 4 Hz. We surmise that increased attention to stimulus competition may be required to observe the attentional sampling rhythms we report here. Indeed outside of attended cue periods we did not observe periodic behavioral responses (Figure 3—figure supplement 1), suggesting that attentional sampling is commensurate with sustained and goal-directed attention, instead of persisting throughout rivalry. We also note that the issue of trial-to-trial variability when reporting on perceptual changes cannot be completely avoided in binocular rivalry research, and is important to consider. Here, one might argue that variable timing in perceptual reports may blur any effects of temporal periodicity. However, our results clearly demonstrate that a change in perceptual state occurred for attended low-frequency cues that were unique in changing visual consciousness compared to visual-only baseline, at or near frequencies of attentional sampling that have been reported in previous literature. By comparison, the time-course of visual switches during visual-only periods did not exhibit periodic sampling, nor did the first-switches during non-attended, or high-frequency cues.

Distinct neural correlates of these attentional sampling rhythms were also found in the EEG. We found significantly greater 3.5 Hz ITPC strength for mismatched compared to matched cue types over right fronto-centro-temporal electrodes [FT8 and C6], suggesting this region may be a candidate neural correlate for divided periodic attentional sampling (Figure 4a). Accordingly, following both mismatched and matched cues, increased 3.5 Hz ITPC in this region also positively correlated with the likelihood of switching to the previously suppressed visual image across subjects (Figure 4c–d). Using visual-only stimulation, previous research has identified a pre-target ~4 Hz phase-dependency for peri-threshold perception when attention is divided across visual hemifields (Landau et al., 2015). While our right fronto-temporal region is different to those previously implicated in attentional-sampling (e.g. Landau et al., 2015), we note that in our paradigm, attention was not divided between visual hemifields, but between competing stimuli during binocular rivalry. Right fronto-temporal regions have previously been implicated in the reorientation of attention to unattended locations (Corbetta and Shulman, 2002; Downar et al., 2000; Proskovec et al., 2018), and most recently, Helfrich et al., 2018 have demonstrated that the phase of ~4 Hz dependent sampling in frontal and parietal areas determines visual perception. Taken together, our results support previous research indicating that a periodic attentional sampling mechanism modulates visual perception, here extending this finding into binocular rivalry when visual stimuli spatially overlap and compete for perceptual dominance.

We also found behavioral and neural correlates of focused attentional sampling during binocular rivalry when cues were consistent with the prevailing visual percept. Specifically, 8 Hz ITPC over parieto-occipital electrodes was negatively correlated with the likelihood of switching to the incongruent perceptual outcome (Figure 5d). Previously, phase-dependent peri-threshold perception has been reported for focused attention tasks in the visual domain (Busch et al., 2009; Busch and VanRullen, 2010; Hanslmayr et al., 2013; Mathewson et al., 2009), and has primarily implicated an approximately 7 Hz component located over fronto-central electrodes (Busch et al., 2009; Busch and VanRullen, 2010). Given the differences between paradigms, it is unsurprising that our identified region for focused attentional sampling does not coincide with those reported in previous research regarding phase-dependent perception. Particularly as our right-lateralized response may be due to the left-lateralized tactile input used to investigate crossmodal attentional sampling (though ITPC was not different among the three crossmodal stimulation types, data not shown). While promising, future experiments that control for this lateralization are needed to characterize the contributions of fronto-centro-temporal and parieto-occipital regions to this effect, particularly as activity over each of these regions has previously been implicated in the reorienting of visuo-spatial attention (Corbetta and Shulman, 2002; Downar et al., 2000; Dugué et al., 2018; Proskovec et al., 2018), and for the integration of multisensory stimuli into a coherent percept (Beauchamp, 2005; Bushara et al., 2003; Calvert and Thesen, 2004; Driver and Noesselt, 2008; Zhang et al., 2011a). Increases in right parieto-occipital theta power (4–8 Hz) have also been shown when attending to visual stimuli in the presence of auditory distractors (van Driel et al., 2014), with the phase of right parieto-occipital alpha (8–10 Hz) or theta (6–7 Hz) oscillations determining the perceptual outcome of bistable stimuli (Ronconi et al., 2017). As such, the present modulation for 8 Hz parieto-occipital ITPC is consistent with the idea that right-parietal networks may preferentially represent temporal information in the visual modality (Battelli et al., 2007; Guggisberg et al., 2011).

Reporting binocular rivalry switches involves both a change in perception, and a decision to press the button. Accordingly, the attentional sampling in binocular rivalry we report here may reflect the fluctuation of perception or of decision criterion. Recent studies of behavioral oscillations that have employed signal detection theory have reported that sensitivity and response criterion both exhibit oscillations (at distinct frequencies) in the high theta/low alpha band, for both vision (Zhang et al., 2018) and audition (Ho et al., 2017). Consequently, whether our oscillations reflect perceptual or decision-level effects must be clarified (Ho et al., 2017; Iemi and Busch, 2018; Iemi et al., 2017; Limbach and Corballis, 2016). Our paradigm cannot resolve this as it is, although a future investigation using our paradigm combined with signal detection theory could do so.

Our analysis has so far revealed that when crossmodal cues mismatched with the dominant binocular rivalry stimulus, that rates of attentional sampling slowed to ~3.5 Hz – implicating the division of attention over multiple locations. However, our exogenous cues oriented attention toward the congruency of visual and crossmodal stimuli, prompting the question: between what was attentional sampling divided? One possibility is that attentional sampling during mismatched cues was divided between two sensory modalities, as the brain tried to resolve a conflict between concurrent auditory/tactile and visual information. Figure 6a provides a schematic of this multisensory interpretation. If the neural activity in our identified region is representative of divided sampling between modalities, it constitutes the first evidence that an attentional sampling mechanism can flexibly orient between temporally co-modulating crossmodal stimuli. Although the facilitative role of attention in multisensory integration remains controversial (Hartcher-O’Brien et al., 2016; Talsma et al., 2010), we see it as a viable possibility that this mechanism resolved perceptual ambiguity through a visual perceptual switch to the competing image, rendering the multisensory stimuli congruent.

Figure 6

Download asset Open asset

As only attended, low-frequency modulated cues enabled a change in visual consciousness, we must consider whether the lack of a high-frequency effect reflects an upper limit in temporal frequency on crossmodal interactions or attention. Such a limit on crossmodal interactions may explain why we observed low- but not high-frequency behavioral effects in the present task, and is supported by previous investigations regarding the binding of multisensory stimulus attributes (Fujisaki and Nishida, 2005; Fujisaki and Nishida, 2010; Lunghi et al., 2014; Vroomen and Keetels, 2010), and the limits of crossmodal temporal judgments (Fujisaki and Nishida, 2005; Fujisaki and Nishida, 2010; Holcombe, 2009; Vroomen and Keetels, 2010). For example, Fujisaki and Nishida, 2005 have shown that judgments of temporal synchrony between rhythmic sensory streams degrade above ~4 Hz. It is plausible that the ineffective crossmodal cueing that we found is related to the above-mentioned findings, reflecting a limit on crossmodal integration processes, rather than attention.

Having said that, one previous study using a similar design to ours was successful in eliciting a high-frequency crossmodal effect (15–20 Hz; Lunghi et al., 2014), and notably, in the absence of explicit attentional demands. However, these differences are not wholly unexpected, as to optimize the present task for EEG recordings we used larger (6.5° visual angle) luminance-modulated sinusoidal gratings to facilitate subsequent steady-state visually evoked potential analyses. While in comparison, Lunghi et al. succeeded in showing a high-frequency effect with rivalry stimuli that were contrast-modulated narrow-band random noise patterns (3.2° visual angle), and did so under conditions analogous to our non-attend conditions. This difference in the composition of visual stimuli is noteworthy, as stimulus size is known to strongly affect rivalry dynamics (Blake et al., 1992). To our knowledge, whether stimulus size impacts upon crossmodal effects during binocular rivalry is unknown. However, given the strength of our results for attended low-frequency flicker (Figure 2a), we note that the low- and high-frequency effects observed by Lunghi et al., 2014 are not generalizable to the larger and attended-rivalry stimuli employed here. Similarly, whether the type of stimuli (e.g., gratings vs random noise patterns) also impacts upon crossmodal effects during rivalry represents a fruitful endeavor for research, particularly given the novel possibility of distinguishing between crossmodal and attentional limits on attentional sampling.

An alternate possibility to crossmodal attentional sampling is that the 3.5 Hz rhythm in our paradigm reflects divided attentional sampling between dominant and suppressed visual images during binocular rivalry (Figure 6b). The frequency of divided attentional sampling that we observed is consistent with those obtained when visual attention has been divided between two objects or locations (Fiebelkorn et al., 2013; Landau and Fries, 2012). As our binocular rivalry stimuli necessarily occupied the same spatial location, attention in our paradigm was likely divided between either features or objects, instead of locations. Indeed, feature-based attention has already been shown to modulate neural processes when an attended target is suppressed during continuous flash suppression (Kanai et al., 2006). During binocular rivalry, perceptual dominance is also influenced by object-based attention (Mitchell et al., 2004), with unconscious selection mechanisms argued to mediate perceptual transitions (Lin and He, 2009). This second alternative is also indirectly supported by the temporal limits of binocular rivalry when conflicting visual stimuli are presented asynchronously, without temporal overlap between the two eyes (O'Shea and Blake, 1986; van Boxtel et al., 2008b; van Boxtel et al., 2007). The maximum stimulus onset asynchrony that can sustain this type of rivalry is approximately 350 ± 50 ms, beyond which alternating stimuli introduced to one eye are perceived immediately, without rivalry occurring (van Boxtel et al., 2008a). This limit is consistent with a 7–8 Hz attentional sampling rhythm distributed between the two conflicting stimuli (each sampled at ~3–4 Hz). When stimuli are presented rapidly enough they are temporally bound together and can engage in ongoing rivalry; when stimuli are presented slower than at 3–4 Hz, they are temporally individuated by attention, and rivalry ceases. A recent computational model that explicitly modeled time-varying attention could indeed reproduce this finding (Li et al., 2017), suggesting that attention is the process that temporally binds the successive stimulus presentations together. Based on our findings, we propose the persistence of percepts may also be modeled using an oscillatory, and feature-selective attentional mechanism (Li et al., 2015; Li et al., 2017).

The suggestion that attention can sample between conscious and nonconscious vision is also consistent with a view that the underlying neuronal processes for attention and consciousness are supported by distinct neural mechanisms (Bahrami et al., 2007; Watanabe et al., 2011); for review see Koch and Tsuchiya, 2007). We note that while attentional sampling of a suppressed image suggests that attention is not sufficient for consciousness (Dehaene et al., 2006; Koch and Tsuchiya, 2007; Lamme, 2003; van Boxtel et al., 2010a; van Boxtel et al., 2010b), this interpretation remains consistent with a view that attention may still be necessary for conscious perception (Chica and Bartolomeo, 2012; Cohen and Dennett, 2011; Merikle and Joordens, 1997; O'Regan and Noë, 2001; Posner, 1994; Posner, 2012).

Whether attributable to conscious-nonconscious, or visual-crossmodal attentional sampling, the present results also complement the ‘active-sensing’ hypothesis (Schroeder et al., 2010), whereby perceptual selection is determined by routine exploratory behaviors. According to the active-sensing hypothesis, attention is critical to search for task-relevant information from the environment (Schroeder et al., 2010), particularly via the rhythmic coordination of multisensory information (Schroeder et al., 2010; Thorne and Debener, 2014). Intriguingly, early contributions from multi-sensory (non-visual) information have been shown to modulate perception (Morillon et al., 2014; Schroeder et al., 2010; van Atteveldt et al., 2014). The rhythmic modulation of visual performance has also been demonstrated to follow the onset of both voluntary (Hogendoorn, 2016), and preparatory motor behaviors (Tomassini et al., 2017; Tomassini et al., 2015). Here, in further support of the active-sensing hypothesis, we have shown that task-relevant crossmodal information can change the rhythmic modulations of perceptual selection during competition for perceptual dominance.

In summary, here we have provided novel evidence in support of attentional sampling during binocular rivalry through the use of crossmodal cues matched to either a conscious or nonconscious visual stimulus. As the attention sampling hypothesis continues to garner traction from various psychophysical and neuronal paradigms (Fiebelkorn et al., 2018; Helfrich et al., 2018; VanRullen, 2016a; VanRullen, 2016b; VanRullen, 2018), future targeted experimentation can confirm whether attention can indeed sample across modalities (Figure 6a), as well as if attention can sample between conscious and nonconscious neural representations during binocular rivalry (Figure 6b). The interactions between crossmodal stimuli and conscious perception represent a fruitful avenue for experimentation (Faivre et al., 2017), here uncovering the previously unknown dependence of attention and consciousness on rhythmic neural dynamics of the human brain.

The raw data in this study are available via the Monash University Figshare repository (https://figshare.com/projects/Crossmodal_binocular_rivalry_attention_sampling_project/56252). Analysis code is available via GitHub (https://github.com/Davidson-MJ/BRproject-attentionsampling; copy archived at https://github.com/elifesciences-publications/BRproject-attentionsampling).

1. 1. Matthew J Davidson

   (2018) figshare

   Crossmodal binocular rivalry, attention sampling project.

   https://figshare.com/projects/Crossmodal_binocular_rivalry_attention_sampling_project/56252

1. Book

   1. Alais D
   2. Blake R

   (2005)

   Binocular rivalry

   MIT press.

   • Google Scholar
2. 1. Alais D

   (2012) Binocular rivalry: competition and inhibition in visual perception

   Wiley Interdisciplinary Reviews: Cognitive Science 3:87–103.

   https://doi.org/10.1002/wcs.151

   • PubMed
   • Google Scholar
3. 1. Bahrami B
   2. Lavie N
   3. Rees G

   (2007) Attentional load modulates responses of human primary visual cortex to invisible stimuli

   Current Biology 17:509–513.

   https://doi.org/10.1016/j.cub.2007.01.070

   • PubMed
   • Google Scholar
4. 1. Bastos AM
   2. Schoffelen JM

   (2015) A Tutorial Review of Functional Connectivity Analysis Methods and Their Interpretational Pitfalls

   Frontiers in Systems Neuroscience 9:175.

   https://doi.org/10.3389/fnsys.2015.00175

   • PubMed
   • Google Scholar
5. 1. Battelli L
   2. Pascual-Leone A
   3. Cavanagh P

   (2007) The 'when' pathway of the right parietal lobe

   Trends in Cognitive Sciences 11:204–210.

   https://doi.org/10.1016/j.tics.2007.03.001

   • PubMed
   • Google Scholar
6. 1. Beauchamp MS

   (2005) See me, hear me, touch me: multisensory integration in lateral occipital-temporal cortex

   Current Opinion in Neurobiology 15:145–153.

   https://doi.org/10.1016/j.conb.2005.03.011

   • PubMed
   • Google Scholar
7. 1. Benjamin DJ
   2. Berger JO
   3. Johannesson M
   4. Nosek BA
   5. Wagenmakers E-J
   6. Berk R
   7. Bollen KA
   8. Brembs B
   9. Brown L
   10. Camerer C
   11. Cesarini D
   12. Chambers CD
   13. Clyde M
   14. Cook TD
   15. De Boeck P
   16. Dienes Z
   17. Dreber A
   18. Easwaran K
   19. Efferson C
   20. Fehr E
   21. Fidler F
   22. Field AP
   23. Forster M
   24. George EI
   25. Gonzalez R
   26. Goodman S
   27. Green E
   28. Green DP
   29. Greenwald AG
   30. Hadfield JD
   31. Hedges LV
   32. Held L
   33. Hua Ho T
   34. Hoijtink H
   35. Hruschka DJ
   36. Imai K
   37. Imbens G
   38. Ioannidis JPA
   39. Jeon M
   40. Jones JH
   41. Kirchler M
   42. Laibson D
   43. List J
   44. Little R
   45. Lupia A
   46. Machery E
   47. Maxwell SE
   48. McCarthy M
   49. Moore DA
   50. Morgan SL
   51. Munafó M
   52. Nakagawa S
   53. Nyhan B
   54. Parker TH
   55. Pericchi L
   56. Perugini M
   57. Rouder J
   58. Rousseau J
   59. Savalei V
   60. Schönbrodt FD
   61. Sellke T
   62. Sinclair B
   63. Tingley D
   64. Van Zandt T
   65. Vazire S
   66. Watts DJ
   67. Winship C
   68. Wolpert RL
   69. Xie Y
   70. Young C
   71. Zinman J
   72. Johnson VE

   (2018) Redefine statistical significance

   Nature Human Behaviour 2:6–10.

   https://doi.org/10.1038/s41562-017-0189-z

   • Google Scholar
8. 1. Benjamini Y
   2. Yekutieli D

   (2001) The Control of the False Discovery Rate in Multiple Testing under Dependency

   The Annals of Statistics 29:1165–1188.

   https://doi.org/10.1214/aos/1013699998

   • Google Scholar
9. 1. Blake R
   2. O'Shea RP
   3. Mueller TJ

   (1992) Spatial zones of binocular rivalry in central and peripheral vision

   Visual Neuroscience 8:469–478.

   https://doi.org/10.1017/S0952523800004971

   • PubMed
   • Google Scholar
10. 1. Bokil H
    2. Andrews P
    3. Kulkarni JE
    4. Mehta S
    5. Mitra PP

    (2010) Chronux: a platform for analyzing neural signals

    Journal of Neuroscience Methods 192:146–151.

    https://doi.org/10.1016/j.jneumeth.2010.06.020

    • PubMed
    • Google Scholar
11. 1. Brainard DH

    (1997) The Psychophysics Toolbox

    Spatial Vision 10:433–436.

    https://doi.org/10.1163/156856897X00357

    • PubMed
    • Google Scholar
12. 1. Busch NA
    2. Dubois J
    3. VanRullen R

    (2009) The phase of ongoing EEG oscillations predicts visual perception

    Journal of Neuroscience 29:7869–7876.

    https://doi.org/10.1523/JNEUROSCI.0113-09.2009

    • PubMed
    • Google Scholar
13. 1. Busch NA
    2. VanRullen R

    (2010) Spontaneous EEG oscillations reveal periodic sampling of visual attention

    PNAS 107:16048–16053.

    https://doi.org/10.1073/pnas.1004801107

    • PubMed
    • Google Scholar
14. 1. Bushara KO
    2. Hanakawa T
    3. Immisch I
    4. Toma K
    5. Kansaku K
    6. Hallett M

    (2003) Neural correlates of cross-modal binding

    Nature Neuroscience 6:190–195.

    https://doi.org/10.1038/nn993

    • PubMed
    • Google Scholar
15. 1. Calvert GA
    2. Thesen T

    (2004) Multisensory integration: methodological approaches and emerging principles in the human brain

    Journal of Physiology-Paris 98:191–205.

    https://doi.org/10.1016/j.jphysparis.2004.03.018

    • PubMed
    • Google Scholar
16. 1. Chakravarthi R
    2. Vanrullen R

    (2012) Conscious updating is a rhythmic process

    PNAS 109:10599–10604.

    https://doi.org/10.1073/pnas.1121622109

    • PubMed
    • Google Scholar
17. 1. Chen A
    2. Wang A
    3. Wang T
    4. Tang X
    5. Zhang M

    (2017) Behavioral Oscillations in Visual Attention Modulated by Task Difficulty

    Frontiers in Psychology 8:1630.

    https://doi.org/10.3389/fpsyg.2017.01630

    • PubMed
    • Google Scholar
18. 1. Chica AB
    2. Bartolomeo P

    (2012) Attentional routes to conscious perception

    Frontiers in Psychology 3:1.

    https://doi.org/10.3389/fpsyg.2012.00001

    • PubMed
    • Google Scholar
19. 1. Chong SC
    2. Tadin D
    3. Blake R

    (2005) Endogenous attention prolongs dominance durations in binocular rivalry

    Journal of Vision 5:6.

    https://doi.org/10.1167/5.11.6

    • Google Scholar
20. 1. Chong SC
    2. Blake R

    (2006) Exogenous attention and endogenous attention influence initial dominance in binocular rivalry

    Vision Research 46:1794–1803.

    https://doi.org/10.1016/j.visres.2005.10.031

    • PubMed
    • Google Scholar
21. 1. Chopin A
    2. Mamassian P

    (2010) Task usefulness affects perception of rivalrous images

    Psychological Science 21:1886–1893.

    https://doi.org/10.1177/0956797610389190

    • PubMed
    • Google Scholar
22. 1. Cohen MA
    2. Dennett DC

    (2011) Consciousness cannot be separated from function

    Trends in Cognitive Sciences 15:358–364.

    https://doi.org/10.1016/j.tics.2011.06.008

    • PubMed
    • Google Scholar
23. 1. Conrad V
    2. Bartels A
    3. Kleiner M
    4. Noppeney U

    (2010) Audiovisual interactions in binocular rivalry

    Journal of Vision 10:27.

    https://doi.org/10.1167/10.10.27

    • PubMed
    • Google Scholar
24. 1. Corbetta M
    2. Shulman GL

    (2002) Control of goal-directed and stimulus-driven attention in the brain

    Nature Reviews Neuroscience 3:201–215.

    https://doi.org/10.1038/nrn755

    • PubMed
    • Google Scholar
25. Software

    1. Davidson MJ

    (2018) BRproject-attentionsampling, version dfa6de4

    GitHub.

    https://github.com/Davidson-MJ/BRproject-attentionsampling
26. 1. Dehaene S
    2. Changeux JP
    3. Naccache L
    4. Sackur J
    5. Sergent C

    (2006) Conscious, preconscious, and subliminal processing: a testable taxonomy

    Trends in Cognitive Sciences 10:204–211.

    https://doi.org/10.1016/j.tics.2006.03.007

    • PubMed
    • Google Scholar
27. 1. Deroy O
    2. Chen Y-C
    3. Spence C

    (2014) Multisensory constraints on awareness

    Philosophical Transactions of the Royal Society B: Biological Sciences 369:20130207.

    https://doi.org/10.1098/rstb.2013.0207

    • Google Scholar
28. 1. Deroy O
    2. Faivre N
    3. Lunghi C
    4. Spence C
    5. Aller M
    6. Noppeney U

    (2016) The Complex Interplay Between Multisensory Integration and Perceptual Awareness

    Multisensory Research 29:585–606.

    https://doi.org/10.1163/22134808-00002529

    • PubMed
    • Google Scholar
29. 1. Dieter KC
    2. Tadin D

    (2011) Understanding attentional modulation of binocular rivalry: a framework based on biased competition

    Frontiers in Human Neuroscience 5:155.

    https://doi.org/10.3389/fnhum.2011.00155

    • PubMed
    • Google Scholar
30. 1. Dieter KC
    2. Melnick MD
    3. Tadin D

    (2015) When can attention influence binocular rivalry?

    Attention, Perception, & Psychophysics 77:1908–1918.

    https://doi.org/10.3758/s13414-015-0905-6

    • PubMed
    • Google Scholar
31. 1. Dieter KC
    2. Brascamp J
    3. Tadin D
    4. Blake R

    (2016a) Does visual attention drive the dynamics of bistable perception?

    Attention, Perception, & Psychophysics 78:1861–1873.

    https://doi.org/10.3758/s13414-016-1143-2

    • PubMed
    • Google Scholar
32. 1. Dieter KC
    2. Melnick MD
    3. Tadin D

    (2016b) Perceptual training profoundly alters binocular rivalry through both sensory and attentional enhancements

    PNAS 113:12874–12879.

    https://doi.org/10.1073/pnas.1602722113

    • PubMed
    • Google Scholar
33. 1. Downar J
    2. Crawley AP
    3. Mikulis DJ
    4. Davis KD

    (2000) A multimodal cortical network for the detection of changes in the sensory environment

    Nature Neuroscience 3:277–283.

    https://doi.org/10.1038/72991

    • PubMed
    • Google Scholar
34. 1. Driver J
    2. Noesselt T

    (2008) Multisensory interplay reveals crossmodal influences on 'sensory-specific' brain regions, neural responses, and judgments

    Neuron 57:11–23.

    https://doi.org/10.1016/j.neuron.2007.12.013

    • PubMed
    • Google Scholar
35. 1. Dugué L
    2. Vanrullen R

    (2014) The dynamics of attentional sampling during visual search revealed by Fourier analysis of periodic noise interference

    Journal of Vision 14:11–15.

    https://doi.org/10.1167/14.2.11

    • PubMed
    • Google Scholar
36. 1. Dugué L
    2. McLelland D
    3. Lajous M
    4. VanRullen R

    (2015) Attention searches nonuniformly in space and in time

    PNAS 112:15214–15219.

    https://doi.org/10.1073/pnas.1511331112

    • PubMed
    • Google Scholar
37. 1. Dugué L
    2. Roberts M
    3. Carrasco M

    (2016) Attention reorients periodically

    Current Biology 26:1595–1601.

    https://doi.org/10.1016/j.cub.2016.04.046

    • PubMed
    • Google Scholar
38. 1. Dugué L
    2. VanRullen R

    (2017a) Transcranial Magnetic Stimulation Reveals Intrinsic Perceptual and Attentional Rhythms

    Frontiers in Neuroscience 11:154.

    https://doi.org/10.3389/fnins.2017.00154

    • PubMed
    • Google Scholar
39. 1. Dugué L
    2. Xue AM
    3. Carrasco M

    (2017b) Distinct perceptual rhythms for feature and conjunction searches

    Journal of Vision 17:22.

    https://doi.org/10.1167/17.3.22

    • PubMed
    • Google Scholar
40. 1. Dugué L
    2. Merriam EP
    3. Heeger DJ
    4. Carrasco M

    (2018) Specific Visual Subregions of TPJ Mediate Reorienting of Spatial Attention

    Cerebral Cortex 28:2375–2390.

    https://doi.org/10.1093/cercor/bhx140

    • PubMed
    • Google Scholar
41. 1. Faivre N
    2. Arzi A
    3. Lunghi C
    4. Salomon R

    (2017) Consciousness is more than meets the eye: a call for a multisensory study of subjective experience†

    Neuroscience of Consciousness 2017:1–8.

    https://doi.org/10.1093/nc/nix003

    • Google Scholar
42. 1. Fiebelkorn IC
    2. Saalmann YB
    3. Kastner S

    (2013) Rhythmic sampling within and between objects despite sustained attention at a cued location

    Current Biology 23:2553–2558.

    https://doi.org/10.1016/j.cub.2013.10.063

    • PubMed
    • Google Scholar
43. 1. Fiebelkorn IC
    2. Pinsk MA
    3. Kastner S

    (2018) A Dynamic interplay within the frontoparietal network underlies rhythmic spatial attention

    Neuron 99:842–853.

    https://doi.org/10.1016/j.neuron.2018.07.038

    • PubMed
    • Google Scholar
44. 1. Frey JN
    2. Ruhnau P
    3. Weisz N

    (2015) Not so different after all: The same oscillatory processes support different types of attention

    Brain Research 1626:183–197.

    https://doi.org/10.1016/j.brainres.2015.02.017

    • PubMed
    • Google Scholar
45. 1. Fries P

    (2015) Rhythms for Cognition: Communication through Coherence

    Neuron 88:220–235.

    https://doi.org/10.1016/j.neuron.2015.09.034

    • PubMed
    • Google Scholar
46. 1. Fujisaki W
    2. Nishida S

    (2005) Temporal frequency characteristics of synchrony-asynchrony discrimination of audio-visual signals

    Experimental Brain Research 166:455–464.

    https://doi.org/10.1007/s00221-005-2385-8

    • PubMed
    • Google Scholar
47. 1. Fujisaki W
    2. Nishida S

    (2010) A common perceptual temporal limit of binding synchronous inputs across different sensory attributes and modalities

    Proceedings of the Royal Society B: Biological Sciences 277:2281–2290.

    https://doi.org/10.1098/rspb.2010.0243

    • Google Scholar
48. 1. Guggisberg AG
    2. Dalal SS
    3. Schnider A
    4. Nagarajan SS

    (2011) The neural basis of event-time introspection

    Consciousness and Cognition 20:1899–1915.

    https://doi.org/10.1016/j.concog.2011.03.008

    • PubMed
    • Google Scholar
49. 1. Guzman-Martinez E
    2. Ortega L
    3. Grabowecky M
    4. Mossbridge J
    5. Suzuki S

    (2012) Interactive coding of visual spatial frequency and auditory amplitude-modulation rate

    Current Biology 22:383–388.

    https://doi.org/10.1016/j.cub.2012.01.004

    • PubMed
    • Google Scholar
50. 1. Hanslmayr S
    2. Volberg G
    3. Wimber M
    4. Dalal SS
    5. Greenlee MW

    (2013) Prestimulus oscillatory phase at 7 Hz gates cortical information flow and visual perception

    Current Biology 23:2273–2278.

    https://doi.org/10.1016/j.cub.2013.09.020

    • PubMed
    • Google Scholar
51. 1. Hartcher-O’Brien J
    2. Talsma D
    3. Adam R
    4. Vercillo T
    5. Macaluso E
    6. Noppeney U

    (2016) The Curious Incident of Attention in Multisensory Integration: Bottom-up vs. Top-down

    Multisensory Research 29:557–583.

    https://doi.org/10.1163/22134808-00002528

    • Google Scholar
52. 1. Helfrich RF
    2. Fiebelkorn IC
    3. Szczepanski SM
    4. Lin JJ
    5. Parvizi J
    6. Knight RT
    7. Kastner S

    (2018) Neural Mechanisms of Sustained Attention Are Rhythmic

    Neuron 99:854–865.

    https://doi.org/10.1016/j.neuron.2018.07.032

    • PubMed
    • Google Scholar
53. 1. Ho HT
    2. Leung J
    3. Burr DC
    4. Alais D
    5. Morrone MC

    (2017) Auditory Sensitivity and Decision Criteria Oscillate at Different Frequencies Separately for the Two Ears

    Current Biology : CB 27:3643–3649.

    https://doi.org/10.1016/j.cub.2017.10.017

    • PubMed
    • Google Scholar
54. 1. Hogendoorn H

    (2016) Voluntary Saccadic Eye Movements Ride the Attentional Rhythm

    Journal of Cognitive Neuroscience 28:1625–1635.

    https://doi.org/10.1162/jocn_a_00986

    • PubMed
    • Google Scholar
55. 1. Holcombe AO

    (2009) Seeing slow and seeing fast: two limits on perception

    Trends in Cognitive Sciences 13:216–221.

    https://doi.org/10.1016/j.tics.2009.02.005

    • PubMed
    • Google Scholar
56. 1. Holcombe AO
    2. Chen WY

    (2013) Splitting attention reduces temporal resolution from 7 Hz for tracking one object to <3 Hz when tracking three

    Journal of Vision 13:12–19.

    https://doi.org/10.1167/13.1.12

    • PubMed
    • Google Scholar
57. 1. Huang Y
    2. Chen L
    3. Luo H

    (2015) Behavioral oscillation in priming: competing perceptual predictions conveyed in alternating theta-band rhythms

    Journal of Neuroscience 35:2830–2837.

    https://doi.org/10.1523/JNEUROSCI.4294-14.2015

    • PubMed
    • Google Scholar
58. 1. Iemi L
    2. Chaumon M
    3. Crouzet SM
    4. Busch NA

    (2017) Spontaneous Neural Oscillations Bias Perception by Modulating Baseline Excitability

    The Journal of Neuroscience 37:807–819.

    https://doi.org/10.1523/JNEUROSCI.1432-16.2016

    • PubMed
    • Google Scholar
59. 1. Iemi L
    2. Busch NA

    (2018) Moment-to-Moment fluctuations in neuronal excitability bias subjective perception rather than strategic decision-making

    Eneuro 5:ENEURO.0430-17.2018.

    https://doi.org/10.1523/ENEURO.0430-17.2018

    • PubMed
    • Google Scholar
60. 1. Jack BN
    2. Hacker G

    (2014) Predictive coding explains auditory and tactile influences on vision during binocular rivalry

    Journal of Neuroscience 34:6423–6424.

    https://doi.org/10.1523/JNEUROSCI.1040-14.2014

    • PubMed
    • Google Scholar
61. 1. Kanai R
    2. Tsuchiya N
    3. Verstraten FA

    (2006) The scope and limits of top-down attention in unconscious visual processing

    Current Biology 16:2332–2336.

    https://doi.org/10.1016/j.cub.2006.10.001

    • PubMed
    • Google Scholar
62. 1. Kang M-S
    2. Blake R

    (2005)

    Perceptual synergy between seeing and hearing revealed during binocular rivalry

    Psichologija 32:7–15.

    • Google Scholar
63. 1. Koch C
    2. Tsuchiya N

    (2007) Attention and consciousness: two distinct brain processes

    Trends in Cognitive Sciences 11:16–22.

    https://doi.org/10.1016/j.tics.2006.10.012

    • PubMed
    • Google Scholar
64. 1. Laing CR
    2. Chow CC

    (2002)

    A spiking neuron model for binocular rivalry

    Journal of Computational Neuroscience 12:39–53.

    • PubMed
    • Google Scholar
65. 1. Lakatos P
    2. O'Connell MN
    3. Barczak A
    4. Mills A
    5. Javitt DC
    6. Schroeder CE

    (2009) The leading sense: supramodal control of neurophysiological context by attention

    Neuron 64:419–430.

    https://doi.org/10.1016/j.neuron.2009.10.014

    • PubMed
    • Google Scholar
66. 1. Lamme VAF

    (2003) Why visual attention and awareness are different

    Trends in Cognitive Sciences 7:12–18.

    https://doi.org/10.1016/S1364-6613(02)00013-X

    • Google Scholar
67. 1. Landau AN
    2. Fries P

    (2012) Attention samples stimuli rhythmically

    Current Biology 22:1000–1004.

    https://doi.org/10.1016/j.cub.2012.03.054

    • PubMed
    • Google Scholar
68. 1. Landau AN
    2. Schreyer HM
    3. van Pelt S
    4. Fries P

    (2015) Distributed Attention Is Implemented through Theta-Rhythmic Gamma Modulation

    Current Biology 25:2332–2337.

    https://doi.org/10.1016/j.cub.2015.07.048

    • PubMed
    • Google Scholar
69. 1. Li HH
    2. Carrasco M
    3. Heeger DJ

    (2015) Deconstructing Interocular Suppression: Attention and Divisive Normalization

    PLoS Computational Biology 11:e1004510.

    https://doi.org/10.1371/journal.pcbi.1004510

    • PubMed
    • Google Scholar
70. 1. Li HH
    2. Rankin J
    3. Rinzel J
    4. Carrasco M
    5. Heeger DJ

    (2017) Attention model of binocular rivalry

    PNAS 114:E6192–E6201.

    https://doi.org/10.1073/pnas.1620475114

    • PubMed
    • Google Scholar
71. 1. Limbach K
    2. Corballis PM

    (2016) Prestimulus alpha power influences response criterion in a detection task

    Psychophysiology 53:1154–1164.

    https://doi.org/10.1111/psyp.12666

    • PubMed
    • Google Scholar
72. 1. Lin Z
    2. He S

    (2009) Seeing the invisible: the scope and limits of unconscious processing in binocular rivalry

    Progress in Neurobiology 87:195–211.

    https://doi.org/10.1016/j.pneurobio.2008.09.002

    • PubMed
    • Google Scholar
73. 1. Lunghi C
    2. Binda P
    3. Morrone MC

    (2010) Touch disambiguates rivalrous perception at early stages of visual analysis

    Current Biology 20:R143–R144.

    https://doi.org/10.1016/j.cub.2009.12.015

    • PubMed
    • Google Scholar
74. 1. Lunghi C
    2. Alais D

    (2013) Touch interacts with vision during binocular rivalry with a tight orientation tuning

    PLoS ONE 8:e58754.

    https://doi.org/10.1371/journal.pone.0058754

    • PubMed
    • Google Scholar
75. 1. Lunghi C
    2. Morrone MC
    3. Alais D

    (2014) Auditory and tactile signals combine to influence vision during binocular rivalry

    The Journal of Neuroscience 34:784–792.

    https://doi.org/10.1523/JNEUROSCI.2732-13.2014

    • PubMed
    • Google Scholar
76. 1. Lunghi C
    2. Alais D

    (2015) Congruent tactile stimulation reduces the strength of visual suppression during binocular rivalry

    Scientific Reports 5:09413.

    https://doi.org/10.1038/srep09413

    • Google Scholar
77. 1. Maier A
    2. Panagiotaropoulos TI
    3. Tsuchiya N
    4. Keliris GA

    (2012) Introduction to research topic - binocular rivalry: a gateway to studying consciousness

    Frontiers in Human Neuroscience 6:263.

    https://doi.org/10.3389/fnhum.2012.00263

    • PubMed
    • Google Scholar
78. 1. Maris E
    2. Oostenveld R

    (2007) Nonparametric statistical testing of EEG- and MEG-data

    Journal of Neuroscience Methods 164:177–190.

    https://doi.org/10.1016/j.jneumeth.2007.03.024

    • PubMed
    • Google Scholar
79. 1. Mathewson KE
    2. Gratton G
    3. Fabiani M
    4. Beck DM
    5. Ro T

    (2009) To see or not to see: prestimulus alpha phase predicts visual awareness

    Journal of Neuroscience 29:2725–2732.

    https://doi.org/10.1523/JNEUROSCI.3963-08.2009

    • PubMed
    • Google Scholar
80. 1. Merikle PM
    2. Joordens S

    (1997) Parallels between perception without attention and perception without awareness

    Consciousness and Cognition 6:219–236.

    https://doi.org/10.1006/ccog.1997.0310

    • PubMed
    • Google Scholar
81. 1. Mitchell JF
    2. Stoner GR
    3. Reynolds JH

    (2004) Object-based attention determines dominance in binocular rivalry

    Nature 429:410–413.

    https://doi.org/10.1038/nature02584

    • PubMed
    • Google Scholar
82. 1. Morillon B
    2. Schroeder CE
    3. Wyart V

    (2014) Motor contributions to the temporal precision of auditory attention

    Nature Communications 5:5255.

    https://doi.org/10.1038/ncomms6255

    • PubMed
    • Google Scholar
83. 1. O'Regan JK
    2. Noë A

    (2001) A sensorimotor account of vision and visual consciousness

    Behavioral and Brain Sciences 24:939–973.

    https://doi.org/10.1017/S0140525X01000115

    • PubMed
    • Google Scholar
84. 1. O'Shea RP
    2. Blake R

    (1986) Dichoptic temporal frequency differences do not lead to binocular rivalry

    Perception & Psychophysics 39:59–63.

    https://doi.org/10.3758/BF03207584

    • PubMed
    • Google Scholar
85. 1. Paffen CL
    2. Alais D

    (2011) Attentional modulation of binocular rivalry

    Frontiers in Human Neuroscience 5:105.

    https://doi.org/10.3389/fnhum.2011.00105

    • PubMed
    • Google Scholar
86. 1. Posner MI

    (1994) Attention: the mechanisms of consciousness

    PNAS 91:7398–7403.

    https://doi.org/10.1073/pnas.91.16.7398

    • PubMed
    • Google Scholar
87. 1. Posner MI

    (2012) Attentional networks and consciousness

    Frontiers in Psychology 3:64.

    https://doi.org/10.3389/fpsyg.2012.00064

    • PubMed
    • Google Scholar
88. 1. Proskovec AL
    2. Heinrichs-Graham E
    3. Wiesman AI
    4. McDermott TJ
    5. Wilson TW

    (2018) Oscillatory dynamics in the dorsal and ventral attention networks during the reorienting of attention

    Human Brain Mapping 39:2177–2190.

    https://doi.org/10.1002/hbm.23997

    • PubMed
    • Google Scholar
89. 1. Romei V
    2. Gross J
    3. Thut G

    (2012) Sounds reset rhythms of visual cortex and corresponding human visual perception

    Current Biology 22:807–813.

    https://doi.org/10.1016/j.cub.2012.03.025

    • PubMed
    • Google Scholar
90. 1. Ronconi L
    2. Oosterhof NN
    3. Bonmassar C
    4. Melcher D

    (2017) Multiple oscillatory rhythms determine the temporal organization of perception

    PNAS 114:13435–13440.

    https://doi.org/10.1073/pnas.1714522114

    • PubMed
    • Google Scholar
91. 1. Schroeder CE
    2. Lakatos P

    (2009) Low-frequency neuronal oscillations as instruments of sensory selection

    Trends in Neurosciences 32:9–18.

    https://doi.org/10.1016/j.tins.2008.09.012

    • PubMed
    • Google Scholar
92. 1. Schroeder CE
    2. Wilson DA
    3. Radman T
    4. Scharfman H
    5. Lakatos P

    (2010) Dynamics of Active Sensing and perceptual selection

    Current Opinion in Neurobiology 20:172–176.

    https://doi.org/10.1016/j.conb.2010.02.010

    • PubMed
    • Google Scholar
93. 1. Song K
    2. Meng M
    3. Chen L
    4. Zhou K
    5. Luo H

    (2014) Behavioral oscillations in attention: rhythmic α pulses mediated through θ band

    Journal of Neuroscience 34:4837–4844.

    https://doi.org/10.1523/JNEUROSCI.4856-13.2014

    • PubMed
    • Google Scholar
94. 1. Spaak E
    2. de Lange FP
    3. Jensen O

    (2014) Local entrainment of α oscillations by visual stimuli causes cyclic modulation of perception

    Journal of Neuroscience 34:3536–3544.

    https://doi.org/10.1523/JNEUROSCI.4385-13.2014

    • PubMed
    • Google Scholar
95. 1. Talsma D
    2. Senkowski D
    3. Soto-Faraco S
    4. Woldorff MG

    (2010) The multifaceted interplay between attention and multisensory integration

    Trends in Cognitive Sciences 14:400–410.

    https://doi.org/10.1016/j.tics.2010.06.008

    • PubMed
    • Google Scholar
96. 1. Thorne JD
    2. Debener S

    (2014) Look now and hear what's coming: on the functional role of cross-modal phase reset

    Hearing Research 307:144–152.

    https://doi.org/10.1016/j.heares.2013.07.002

    • PubMed
    • Google Scholar
97. 1. Tomassini A
    2. Spinelli D
    3. Jacono M
    4. Sandini G
    5. Morrone MC

    (2015) Rhythmic oscillations of visual contrast sensitivity synchronized with action

    Journal of Neuroscience 35:7019–7029.

    https://doi.org/10.1523/JNEUROSCI.4568-14.2015

    • PubMed
    • Google Scholar
98. 1. Tomassini A
    2. Ambrogioni L
    3. Medendorp WP
    4. Maris E

    (2017) Theta oscillations locked to intended actions rhythmically modulate perception

    eLife 6:e25618.

    https://doi.org/10.7554/eLife.25618

    • PubMed
    • Google Scholar
99. 1. van Atteveldt N
    2. Murray MM
    3. Thut G
    4. Schroeder CE

    (2014) Multisensory integration: flexible use of general operations

    Neuron 81:1240–1253.

    https://doi.org/10.1016/j.neuron.2014.02.044

    • PubMed
    • Google Scholar
100. 1. van Boxtel JJ
     2. van Ee R
     3. Erkelens CJ

     (2007) Dichoptic masking and binocular rivalry share common perceptual dynamics

     Journal of Vision 7:3.

     https://doi.org/10.1167/7.14.3

     • PubMed
     • Google Scholar
101. 1. van Boxtel JJ
     2. Alais D
     3. Erkelens CJ
     4. van Ee R

     (2008a) The role of temporally coarse form processing during binocular rivalry

     PLoS ONE 3:e1429.

     https://doi.org/10.1371/journal.pone.0001429

     • PubMed
     • Google Scholar
102. 1. van Boxtel JJ
     2. Knapen T
     3. Erkelens CJ
     4. van Ee R

     (2008b) Removal of monocular interactions equates rivalry behavior for monocular, binocular, and stimulus rivalries

     Journal of Vision 8:13.

     https://doi.org/10.1167/8.15.13

     • PubMed
     • Google Scholar
103. 1. van Boxtel JJ
     2. Tsuchiya N
     3. Koch C

     (2010a) Consciousness and attention: on sufficiency and necessity

     Frontiers in Psychology 1:217.

     https://doi.org/10.3389/fpsyg.2010.00217

     • PubMed
     • Google Scholar
104. 1. van Boxtel JJ
     2. Tsuchiya N
     3. Koch C

     (2010b) Opposing effects of attention and consciousness on afterimages

     PNAS 107:8883–8888.

     https://doi.org/10.1073/pnas.0913292107

     • PubMed
     • Google Scholar
105. 1. van Driel J
     2. Knapen T
     3. van Es DM
     4. Cohen MX

     (2014) Interregional alpha-band synchrony supports temporal cross-modal integration

     NeuroImage 101:404–415.

     https://doi.org/10.1016/j.neuroimage.2014.07.022

     • PubMed
     • Google Scholar
106. 1. van Ee R
     2. van Boxtel JJ
     3. Parker AL
     4. Alais D

     (2009) Multisensory congruency as a mechanism for attentional control over perceptual selection

     Journal of Neuroscience 29:11641–11649.

     https://doi.org/10.1523/JNEUROSCI.0873-09.2009

     • PubMed
     • Google Scholar
107. 1. VanRullen R
     2. Carlson T
     3. Cavanagh P

     (2007) The blinking spotlight of attention

     PNAS 104:19204–19209.

     https://doi.org/10.1073/pnas.0707316104

     • PubMed
     • Google Scholar
108. 1. Vanrullen R
     2. Dubois J

     (2011) The psychophysics of brain rhythms

     Frontiers in Psychology 2:203.

     https://doi.org/10.3389/fpsyg.2011.00203

     • PubMed
     • Google Scholar
109. 1. VanRullen R

     (2013) Visual attention: a rhythmic process?

     Current Biology 23:R1110–R1112.

     https://doi.org/10.1016/j.cub.2013.11.006

     • PubMed
     • Google Scholar
110. 1. VanRullen R

     (2016a) Perceptual Cycles

     Trends in Cognitive Sciences 20:723–735.

     https://doi.org/10.1016/j.tics.2016.07.006

     • PubMed
     • Google Scholar
111. Book

     1. VanRullen R

     (2016b)

     Perceptual Rhythms

     In: Wixted JT, Serences JT, editors. Stevens’ Handbook of Experimental Psychology and Cognitive Neuroscience (Fourth Edition). United States: John Wiley and Sons, Inc. pp. 525–568.

     • Google Scholar
112. 1. VanRullen R

     (2018) Attention Cycles

     Neuron 99:632–634.

     https://doi.org/10.1016/j.neuron.2018.08.006

     • PubMed
     • Google Scholar
113. 1. Vroomen J
     2. Keetels M

     (2010) Perception of intersensory synchrony: a tutorial review

     Attention, Perception, & Psychophysics 72:871–884.

     https://doi.org/10.3758/APP.72.4.871

     • PubMed
     • Google Scholar
114. 1. Watanabe M
     2. Cheng K
     3. Murayama Y
     4. Ueno K
     5. Asamizuya T
     6. Tanaka K
     7. Logothetis N

     (2011) Attention but not awareness modulates the BOLD signal in the human V1 during binocular suppression

     Science 334:829–831.

     https://doi.org/10.1126/science.1203161

     • PubMed
     • Google Scholar
115. 1. Wilson HR

     (2003) Computational evidence for a rivalry hierarchy in vision

     PNAS 100:14499–14503.

     https://doi.org/10.1073/pnas.2333622100

     • PubMed
     • Google Scholar
116. 1. Zhang D
     2. Hong B
     3. Gao X
     4. Gao S
     5. Röder B

     (2011a) Exploring steady-state visual evoked potentials as an index for intermodal and crossmodal spatial attention

     Psychophysiology 48:665–675.

     https://doi.org/10.1111/j.1469-8986.2010.01132.x

     • PubMed
     • Google Scholar
117. 1. Zhang P
     2. Jamison K
     3. Engel S
     4. He B
     5. He S

     (2011b) Binocular rivalry requires visual attention

     Neuron 71:362–369.

     https://doi.org/10.1016/j.neuron.2011.05.035

     • PubMed
     • Google Scholar
118. 1. Zhang H
     2. Morrone MC
     3. Alais D.

     (2018)

     Behavioural oscillations in visual orientation discrimination reveal distinct modulation rates for both sensitivity and response bias

     Scientific Reports  in press:.

     • Google Scholar
119. 1. Zoefel B
     2. VanRullen R

     (2017) Oscillatory Mechanisms of Stimulus Processing and Selection in the Visual and Auditory Systems: State-of-the-Art, Speculations and Suggestions

     Frontiers in Neuroscience 11:296.

     https://doi.org/10.3389/fnins.2017.00296

     • PubMed
     • Google Scholar

Thank you for submitting your article "Attention periodically samples competing stimuli during binocular rivalry" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Michael Frank as the Senior Editor. The following individual involved in review of your submission has agreed to reveal his identity: Daniel H Baker (Reviewer #3).

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 is an innovative, interesting and timely study that reveals periodic switching dynamics during binocular rivalry, following cross-modal cues, and explores the rate of attentional sampling during dominance and suppression. Data from perceptual responses show that with relevant cross-modal cues attentional modulations are periodic. When the cue is congruent with the current visual percept, switches display an 8Hz periodicity; when the cue is incongruent, switches occur in about half of 8Hz, with a 3.5Hz frequency. This pattern of results fits well with the idea of attentional sampling, whereby attention samples a single target at around 8Hz, but for two competing targets the effective sampling rate is divided, i.e. about 4Hz. The presence of EEG correlates of these rhythmic dynamics at the same frequencies tends to support the authors' claims.

Addressing the following points will provide information regarding the validity of the claims and strengthen the manuscript.

Essential revisions:

Theoretical issues

1) You should discuss the Dieter, Melnick and Tadin, 2015 study, which shows that stimulus-driven cues can bias perceptual report. Exogenous feature-based cues were presented close to the rivalry stimuli; the cue congruent with the (currently) dominant percept lengthened its duration whereas the cue congruent with the suppressed percept hastened it to emerge back. The authors should stress the novelty of their study regarding the oscillatory nature of such process.

2) The authors should discuss whether they think that the oscillatory behavioral responses they found reflect an oscillation of perception or decision criteria, even though these two aspects cannot be distinguished in the perceptual report used in this study. Germane to this discussion is a recent study that using signal detection theory found that alpha oscillation modulates criterion, not sensitivity (Lemi et al., 2017).

3) The reported effects only occur for the 4.5-Hz modulated stimulus, and not for the 20Hz modulation. Does this imply an upper temporal limit for crossmodal interactions or for attention?

4) The authors should relate their findings to computational models of binocular rivalry. Many use mutual inhibition, adaptation and noise to describe the neural circuit underlying rivalry (e.g. Wilson, 1997; Laing and Chow, 2002). Some have explicitly implemented an attentional component (e.g. Li et al., 2017). None of the rivalry models have attempted to model oscillatory perceptual rhythms. Could the current results inform which model component(s) (e.g. input drive, mutual inhibition or attention) may be oscillatory?

5) The authors suggest that the 7-8 Hz oscillation reflects a single attentional focus, and the ~4Hz oscillation reflects attention that was split into two foci. Does that imply that without a mismatched cue, observers' attention would be predominantly allocated to the dominant image without oscillating between the two stimuli? Were that the case, how would this relate to the theory on rivalry. Are the authors suggesting that a mismatch cue allowed attention to start jumping across two items? And that such a process would not happen without a mismatched cue?

Methods, Analyses and Results

6) It is not clear that there are 3 types of cross-modal cues (auditory, tactile, and both combined) until the Materials and methods section; i.e., well after reading the Results and considering the figures. The existence of 3 cue types should be mentioned explicitly earlier. In addition, this begs the question of possible differences among the cue types: Do they affect the dynamics of switching and the results described in all figures?

7) It would be good to report the frequency tagged responses to the rivaling stimuli themselves. Are there increases in the response when a frequency-matched cross-modal cue is presented? Do the amplitudes correspond to participant button presses consistent with Brown and Norcia, 1997?

8) For computing Fourier amplitude spectrum, the authors analyzed the time window from 0.5 to 2s from the cue onset. However, Figure 3C and 3D also show oscillatory patterns in the first 0.5s time window. Include this time, narrow the window, or explain why this range was not analyzed. Note that previous studies have removed a shorter time window after the cue (e.g. Landau and Fries, 2012; Fiebelkorn et al., 2013).

9) It is unclear whether the authors performed FFT on the averaged (across subjects) time course, or on individual time courses, and then averaged the spectrum across subjects. If the former, the latter should also be reported. It is important to ensure that the effect does not merely result from the averaging procedure.

10) A technical issue regards the EEG ITPC calculation and the logic of upsampling. We agree that it is necessary to equate the number of mismatched and matched cue types for the analysis. However, the strategy of upsampling the lowest-trial-count condition up to the trial number in the highest-trial-count condition is questionable. We are not convinced that the authors' analysis confirmed that upsampling, compared to downsampling, reduced the bias introduced when equating ITPC values across subjects. Running the Matlab code below, the authors will see that the ITPC bias caused by a difference in trial count (first figure) is not reduced when upsampling the low-trial-count condition (second figure), but disappears when downsampling the high-trial-count condition (third figure). You should repeat the analysis with downsampling, ideally multiple times to avoid spurious results.

1.%% Matlab code

2. phases = 2*pi*rand(30,1000);% low trial count condition

3. otherphases = 2*pi*rand(90,1000);% high trial count condition

4. downsampled_otherphases = otherphases(1:30,:);% downsample high trial count

5. upsampled_phases = repmat(phases,3,1);% upsample low trial count

6. ITPC = abs(mean(exp(1i*phases),1));

7. otherITPC = abs(mean(exp(1i*otherphases),1));

8. downsampledITPC = abs(mean(exp(1i*downsampled_otherphases),1));

9. upsampledITPC = abs(mean(exp(1i*upsampled_phases),1));

10. figure;

11. subplot(2,1,1); hist(ITPC); title('Original phases (low N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

12. subplot(2,1,2); hist(otherITPC); title('Original phases (high N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

13. figure;

14. subplot(2,1,1); hist(upsampledITPC); title('Upsampled phases (low N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

15. subplot(2,1,2); hist(otherITPC); title('Original phases (high N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

16. figure;

17. subplot(2,1,1); hist(ITPC); title('Original phases (low N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

18. subplot(2,1,2); hist(downsampledITPC); title('Downsampled phases (high N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

19.% % End Matlab code

Figures

11) Figures 2A, 3A, 4B and 5B require figure legends so that the meaning of the colors is clear without reading the caption.

12) In Figures 4A and 5A it would be helpful to mark all electrode locations with small dots. We assume these plots are thresholded, such that electrodes with non-significant t-values are set to t=0, but this wasn't stated explicitly. Why not plot all t-values and indicate which are significant in some other way (e.g. with superimposed symbols).

13) The 9Hz difference in Figure 5B is somewhat unexpected. This is described as a 'stimulus harmonic', but if so, we wouldn't expect to see a difference between matched and mismatched trials. Do the authors have an explanation for this? Does the 9Hz response also correlate with the PSI measure?

14) The model diagrams in Figure 6 are somewhat confusing. For model (a), do the authors mean that the suppressed image emerges back because attention shifts to the mismatched cue first, and thus the mismatched cue enhances the suppressed image? For model (b), it seems surprising that the presence of the green-crossmodal cue would let the attention focus move from the dominant red image to the suppressed red image, before it moves to the dominant green image…

15) For correlation scatterplots, it would be helpful to include R and p values in the plot. Are the straight lines in these plots regression lines that predict the y-values using the x-values (i.e. minimizing the error in the y-direction)? Given that both of these are dependent variables with presumably no implied causation, we recommend using Deming regression lines instead, which minimize the absolute error between the line and each data point.

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

Thank you for resubmitting your work entitled "Attention periodically samples competing stimuli during binocular rivalry" for further consideration at eLife. Your revised article has been favorably evaluated by Michael Frank (Senior Editor), a Reviewing Editor, and three reviewers.

The three reviewers and I think that you have addressed all the issues raised satisfactorily, but one. The pending issues regards the ITPC calculations and down-sampling vs. up-sampling.

We note that there is a crucial difference in the definition of ITPC "bias": You are talking about minimizing absolute bias (the fact that ITPC is always overestimated when using a finite number of trials, to an extent inversely related to the trial number, e.g. compare Author response image 4D and 4E). Instead, the reviewers are worried about relative bias (the fact that two experimental conditions with equal "true" ITPC may produce very different ITPC values if the number of trials is not equated, e.g. compare Author response image 4A and 4D).

Your upsampling method keeps the absolute bias as low as possible in each condition (given the trial number), but does not reduce relative bias (e.g. compare Author response image 4C and 4D). Only downsampling can remove this relative bias (e.g. compare Author response image 4A and 4E). This comes with the cost of an increase in absolute bias for the high N condition (compare Author response image 4D and 4E), but given that your analysis relies on a comparison of the two conditions, the relative bias must be eradicated; absolute bias changes are just collateral damage.

To summarize:

1) The reviewer made the null assumption that the 2 experimental conditions (matched/mismatched) have the same true ITPC but different numbers of trials. (Here the true ITPC was assumed to be zero, which is the worst-case scenario: no actual phase-locking).

2) When comparing ITPC across the two conditions, a systematic difference was found (Author response image 4A vs. 4D). This difference is necessarily a statistical artifact.

3) The reviewer applied the authors' upsampling method to equate trials; the systematic difference remained (Author response image 4B/4C vs. 4D).

4) The reviewer applied a downsampling method to equate trials; the (artifactual) difference disappeared (Author response image 4A vs. 4E).

5) Conclusion:

The authors are correct that upsampling can preserve the ITPC to have a similar value as that derived from the original dataset (without resampling). But, this is not the point here. The aim of resampling is to let the ITPCs from the two conditions comparable when performing statistical tests. When comparing ITPC from two conditions with unequal trial numbers, upsampling can produce false positives, downsampling will not. Thus, the authors should use downsampling instead of upsampling.

Essential revisions:

Theoretical issues

1) You should discuss the Dieter, Melnick and Tadin, 2015 study, which shows that stimulus-driven cues can bias perceptual report. Exogenous feature-based cues were presented close to the rivalry stimuli; the cue congruent with the (currently) dominant percept lengthened its duration whereas the cue congruent with the suppressed percept hastened it to emerge back. The authors should stress the novelty of their study regarding the oscillatory nature of such process.

We thank the reviewers for this suggestion. We have now cited Dieter et al., 2015, in our Discussion. We have also positioned our findings in support of cue-based effects on rivalry, stressing the novel oscillatory nature of this process (Discussion, third paragraph).

2) The authors should discuss whether they think that the oscillatory behavioral responses they found reflect an oscillation of perception or decision criteria, even though these two aspects cannot be distinguished in the perceptual report used in this study. Germane to this discussion is a recent study that using signal detection theory found that alpha oscillation modulates criterion, not sensitivity (Lemi et al., 2017).

As the reviewers point out our study is unable to tell whether the oscillations in switch rates we have observed reflect changes in perceptual sensitivity or decision criteria. While the results of Iemi et al. support oscillations in decision criteria rather than sensitivity per se (Iemi et al., 2017), others show oscillations in both criterion and sensitivity (Ho et al., 2017; Zhang, Morrone, Alais, Sci Rep, In press) and perception rather than decision criteria (Iemi and Busch, 2018). We have included a short treatment of this distinction in our Discussion (ninth paragraph).

3) The reported effects only occur for the 4.5-Hz modulated stimulus, and not for the 20Hz modulation. Does this imply an upper temporal limit for crossmodal interactions or for attention?

We view it as possible that this difference reflects a temporal limit for crossmodal interactions or for attention. We have clarified our consideration of both alternatives in our Discussion (eleventh paragraph). Although high-frequency effects have been reported in other previous, similar designs (without attention), it is unclear whether our new crossmodal stimulus parameters, or the inclusion of an attentional instruction are responsible for the upper temporal limit we have observed. We believe that distinguishing between these alternatives will be a fruitful line of endeavour in future experimentation, and mention this in our Discussion also.

4) The authors should relate their findings to computational models of binocular rivalry. Many use mutual inhibition, adaptation and noise to describe the neural circuit underlying rivalry (e.g. Wilson, 1997; Laing and Chow, 2002). Some have explicitly implemented an attentional component (e.g. Li et al., 2017). None of the rivalry models have attempted to model oscillatory perceptual rhythms. Could the current results inform which model component(s) (e.g. input drive, mutual inhibition or attention) may be oscillatory?

We thank the reviewers for this suggestion, and have now incorporated a treatment of BR models in our Discussion (third paragraph). We suggest that the attention component of a rivalry model, such as the one that is modelled by Li et al., 2017, can be oscillatory. We have linked our previous discussion of van Boxtel et al., 2008; to an oscillatory model of BR more explicitly (Discussion, twelfth paragraph).

5) The authors suggest that the 7-8 Hz oscillation reflects a single attentional focus, and the ~4Hz oscillation reflects attention that was split into two foci. Does that imply that without a mismatched cue, observers' attention would be predominantly allocated to the dominant image without oscillating between the two stimuli? Were that the case, how would this relate to the theory on rivalry. Are the authors suggesting that a mismatch cue allowed attention to start jumping across two items? And that such a process would not happen without a mismatched cue?

We thank the reviewers for the opportunity to address this clarification, and would like to clarify two points: 1) We do not believe that switches in rivalry are caused only by an attentional sampling mechanism, and 2) nor that attentional switches are completely determined by an 8 / 3.5 Hz rhythm.

As to 1), there are many different types of attention (Koch and Tsuchiya, 2007), and in the context of rivalry, it is unclear the extent to which attention (or type thereof), may mediate perceptual outcome. If anything, both endogenous and exogenous effects of attention are rather limited (Alais and Blake, 2005; Maier, Panagiotaropoulos, Tsuchiya, and Keliris, 2012; Dieter et al., 2015).

As to 2), we view our results to be consistent with previous investigations of attentional sampling. Specifically, that attentional sampling rhythms modulate, but do not completely determine perceptual outcome. We have now clarified this nuance in our Discussion (sixth paragraph).

Methods, Analyses and Results

6) It is not clear that there are 3 types of cross-modal cues (auditory, tactile, and both combined) until the Materials and methods section; i.e., well after reading the Results and considering the figures. The existence of 3 cue types should be mentioned explicitly earlier. In addition, this begs the question of possible differences among the cue types: Do they affect the dynamics of switching and the results described in all figures?

Thank you for this suggestion. We now point out specifically that there were three crossmodal cue types more clearly (subsection “Attending to low-frequency crossmodal stimulation promotes the perceptual dominance of low-frequency flicker during binocular rivalry”, first paragraph).

We found no difference in the crossmodal cue effects based on cue type. That is to say, whether attending to low-frequency auditory, tactile, or combined auditory and tactile cues, the results were the same. We observed an increased probability of perceiving the congruent, low-frequency flicker. A new figure shows these individual time-series, separated by crossmodal modality (Author response image 1). No significant difference between cue types emerged when comparing these time-series (repeated measures ANOVA at each time point, with FDR correction for multiple comparisons).

As the cue differences were not statistically significant, we continued by pooling them for our subsequent behavioural analyses. This had the advantage of increasing the number of trials available per mismatch x match condition, and improved our statistical power. Without pooling across cue types, the number of the trials per cue type are only ~33% of the total number of trials, and not enough to obtain significant results with our spectral analysis for first perceptual switches.

Author response image 1

Download asset Open asset

We also compared the ITPC values obtained at attentional sampling frequencies between our three crossmodal cue types. No significant difference emerged when comparing the spatial topography of attentional sampling frequencies.

7) It would be good to report the frequency tagged responses to the rivaling stimuli themselves. Are there increases in the response when a frequency-matched cross-modal cue is presented? Do the amplitudes correspond to participant button presses consistent with Brown and Norcia, 1997?

We are hesitant to include additional SSVEP, SSAEP, or SSSEP analyses, as the additional analyses and explanation would significantly lengthen the manuscript – and it is already very dense. We would like to reiterate that attentional sampling has been proposed to be related to the optimum phase of sampling rhythms, rather than any modulation in SSVEP power.

Having said that, the answer is ‘yes’, amplitudes correspond with subject button presses consistent with previous research on frequency-tagging during rivalry (e.g. Brown and Norcia, 1997; Zhang et al., 2011; Tononi et al., 1998). We here provide an example of the change in 4.5 Hz frequency-tagged SSVEP strength with button press. The amplitudes correspond with perceptual contents, increasing when perceiving a low-frequency flicker, and decreasing when perceiving the high-frequency flicker. We also provide for the reviewer an example of increased 4.5 Hz SNR, during these perceptual switches, when accompanied by a steady-state crossmodal stimulus (Author response image 2).

Author response image 2

Download asset Open asset

8) For computing Fourier amplitude spectrum, the authors analyzed the time window from 0.5 to 2s from the cue onset. However, Figure 3C and 3D also show oscillatory patterns in the first 0.5s time window. Include this time, narrow the window, or explain why this range was not analyzed. Note that previous studies have removed a shorter time window after the cue (e.g. Landau and Fries, 2012; Fiebelkorn et al., 2013).

Regardless of the specific time windows suggested by the reviewers, our conclusions would not change.

Analyzing the spectra between 0.1 and 2 seconds after cue onset, demonstrates the same results, as shown in Author response image 3. However, we reiterate our two reasons for avoiding this early time-window (0-0.5 seconds after stimulus onset):

1) Manuscript Figure 2A shows that the attentional effects in behavior emerge only beyond 0.5 seconds after cue onset, implying that the effects before 0.5 second are not due to crossmodal attentional effects.

2) Figure 3—figure supplement 1 has been included to show that a transient increase in the rate of first switches occurs within this window (between 0-0.5 s), regardless of attentional effects (i.e., in all cue types), as well as after cue offset, implying that the effects before 0.5 second are not cue dependent. Thus, we chose to avoid this early window and focus on cue-specific effects.

Author response image 3

Download asset Open asset

9) It is unclear whether the authors performed FFT on the averaged (across subjects) time course, or on individual time courses, and then averaged the spectrum across subjects. If the former, the latter should also be reported. It is important to ensure that the effect does not merely result from the averaging procedure.

We would like to state that our evidence for periodic sampling cannot result from our averaging procedure. Although the same averaging procedure was used for all conditions, we observed a differentiation of periodicity only for mismatched vs matched cues, and not unattended, high-frequency, visual-only, or post-offset time-series.

The time-course of first switches we display (manuscript Figure 3) displays the proportion of all first switches after cue onset, at each time point averaged across subjects. Our FFT is performed on this average across subjects time-course.

The pattern in individual subjects is similar, though is not present for each individual. This is because the number of switches per each of 8 conditions (match vs. mismatch x low- vs. high- frequency x attended vs. unattended) is very small per subject, and the attentional effects themselves vary across subjects – as shown in manuscript Figure 2B. Furthermore, some FFT spectra from subjects with very few perceptual switches are extremely noisy, and disrupts any effects that could be observed at the group level if averaging individual spectra.

We have now stated that this effect is clearest in the across-subject time-course, and that this analysis was chosen to match with previous investigations of attentional sampling, which also analyzed the time-course of data at the across-subject level (e.g., Landau and Fries, 2012; Fiebelkorn et al., 2013). These revisions are in the last paragraph of the subsection “Binocular rivalry dynamics during attended low-frequency crossmodal cues”.

10) A technical issue regards the EEG ITPC calculation and the logic of upsampling. We agree that it is necessary to equate the number of mismatched and matched cue types for the analysis. However, the strategy of upsampling the lowest-trial-count condition up to the trial number in the highest-trial-count condition is questionable. We are not convinced that the authors' analysis confirmed that upsampling, compared to downsampling, reduced the bias introduced when equating ITPC values across subjects. Running the Matlab code below, the authors will see that the ITPC bias caused by a difference in trial count (first figure) is not reduced when upsampling the low-trial-count condition (second figure), but disappears when downsampling the high-trial-count condition (third figure). You should repeat the analysis with downsampling, ideally multiple times to avoid spurious results.

1.% % Matlab code

2. phases = 2*pi*rand(30,1000);% low trial count condition

3. otherphases = 2*pi*rand(90,1000);% high trial count condition

4. downsampled_otherphases = otherphases(1:30,:);% downsample high trial count

5. upsampled_phases = repmat(phases,3,1);% upsample low trial count

6. ITPC = abs(mean(exp(1i*phases),1));

7. otherITPC = abs(mean(exp(1i*otherphases),1));

8. downsampledITPC = abs(mean(exp(1i*downsampled_otherphases),1));

9. upsampledITPC = abs(mean(exp(1i*upsampled_phases),1));

10. figure;

11. subplot(2,1,1); hist(ITPC); title('Original phases (low N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

12. subplot(2,1,2); hist(otherITPC); title('Original phases (high N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

13. figure;

14. subplot(2,1,1); hist(upsampledITPC); title('Upsampled phases (low N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

15. subplot(2,1,2); hist(otherITPC); title('Original phases (high N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

16. figure;

17. subplot(2,1,1); hist(ITPC); title('Original phases (low N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

18. subplot(2,1,2); hist(downsampledITPC); title('Downsampled phases (high N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

19.% % End Matlab code

We thank the reviewers for providing Matlab code to compare upsampling and downsampling of ITPC. While the reviewers interpretation of their code appears to contradict with our original claim, in fact, it is consistent with what we have claimed: downsampling enhances the bias while upsampling does not. The discordance in interpretation comes from the way the reviewer contrasts the results (ITPC vs. otherITPC, upsampled ITPC vs. otherITPC, and ITPC vs. downsampled ITPC). As we explain in the following, we claim that the appropriate comparison is between ITPC vs upsampled ITPC and other ITPC vs. downsampled ITPC. The details in response to the reviewers’ code are as follows:

Explanation of the code:

Lines 2-5 prepare the data, in which the true ITPC is assumed as 0. We interpret phases to represent 30 trials sampled from 1000 channels (or subjects). Then, line 10-11 plots the expected “bias” distribution over 1000 channels where the true ITPC is 0 and sampled over 30 trials in each channel. (Reproduced as Author response image 4A)

Now line 13-14 plots what the reviewers call upsampled_phases. But line 5 creates it by duplicating the same random distribution 3 times (Author response image 4B, which is identical to Author response image 4A by definition). This is different from what we did, which is upsampling with replacement. Our procedure would replace line 5 as follows:

for iChannel = 1:1000

upsampled_phasesv2 = phases(randi(30,1, 90), iChannell);

end

With this correction, upsampling minimally affects the distribution (Author response image 4C) from the original distribution (Author response image 4A). This was exactly our point. Upsampling does not introduce further bias into the ITPC estimated from all available 30 trials. From the originally sampled 30 trials, we cannot (and should not be able to) reduce the bias no matter how much upsampling we do.

Next, line 12 plots the expected ITPC distribution of otherphases defined in line 3, where now each channel is sampled with 90 trials. This proves the point that ITPC bias (when the true ITPC is 0) reduces as we increase the number of sample trials. (Reproduced as Author response image 4D)

Line 4 downsamples 30 trials out of the 90 available trials in otherphases. Line 18 plots this downsampled distribution of the ITPC, which proves our point that downsampling introduces further bias (comparing 4D to 4E). This is the reason why we have not used downsampling.

The above code by the reviewer establishes that when the true ITPC is assumed to be 0, 1) we should sample more trials to reduce the bias in ITPC, 2) upsampling with replacement does not introduce much bias, and 3) downsampling introduces significant bias.

Given that our data consists of N=34 subjects, whose minimal number of trials is 12 and maximal is 36, and that the mean ITPC across subjects should be computed with an equal number of trials, we opted for upsampling to the maximum number of trials across subjects.

The revised code is reproduced here for convenience:

1.% % Matlab code

2, phases = 2*pi*rand(30,1000);% low trial count condition

3, otherphases = 2*pi*rand(90,1000);% high trial count condition

4, downsampled_otherphases = otherphases(1:30,:);% downsample high trial count

5. upsampled_phases = repmat(phases,3,1);% upsample low trial count

6. upsampled_phasesv2 =zeros(size(upsampled_phases));

7, for iChannel = 1:1000

8. upsampled_phasesv2(:,iChannel) = phases(randi(30,1, 90), iChannel);

9. end

10. ITPC = abs(mean(exp(1i*phases),1));

11. otherITPC = abs(mean(exp(1i*otherphases),1));

12. downsampledITPC = abs(mean(exp(1i*downsampled_otherphases),1));

13. upsampledITPC = abs(mean(exp(1i*upsampled_phases),1));

14. upsampledITPCv2 = abs(mean(exp(1i*upsampled_phasesv2),1));

15. figure;

16. subplot(5,1,1); hist(ITPC); title('Original phases (low N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

17.% figure;

18. subplot(5,1,2); hist(upsampledITPC); title('Upsampled phases (low N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

19. subplot(5,1,3); hist(upsampledITPCv2); title('Upsampled phases v2 (low N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

20. subplot(5,1,4); hist(otherITPC); title('Original phases (high N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

21. subplot(5,1,5); hist(downsampledITPC); title('Downsampled phases (high N)'); set(gca,'xlim',[0 0.4],'ylim',[0 250]); xlabel('ITPC');

22.% % End Matlab code

Author response image 4

Download asset Open asset

As a sanity check, we performed a further analysis using our own data. Author response image 5 shows the results from an example subject, when the number of repetitions is sufficient (100 bootstraps in the example below), the statistical results using:

a) bootstrapping with replacement to retain the original number of trials (N=22 trials for this subject);

b) bootstrapping with replacement when downsampling the number of trials to N=12 (the minimum number across 34 subjects); and

c) bootstrapping with replacement when upsampling the number of trials to N=36 (the maximum across subjects),

All result in similar conclusions.

Critically however, the values for baseline ITPC (or lowest ITPC values) are highest (~0.35) with b) downsampling, compared to upsampling (~0.2) or 22 trials (~0.22), which is consistent with what we expect from the simulation code provided by the reviewer.

Author response image 5

Download asset Open asset

11) Figures 2A, 3A, 4B and 5B require figure legends so that the meaning of the colors is clear without reading the caption.

We have now added these figure legends.

12) In Figures 4A and 5A it would be helpful to mark all electrode locations with small dots. We assume these plots are thresholded, such that electrodes with non-significant t-values are set to t=0, but this wasn't stated explicitly. Why not plot all t-values and indicate which are significant in some other way (e.g. with superimposed symbols).

We had thresholded non-significant electrodes in these figures. We had elected not to superimpose electrodes onto our topoplots that would obscure significant t-scores, and similarly, to threshold all non-significant regions after our two-stage spatial cluster-based analysis. This is because isolated regions (single electrodes) were present in the topography, but did not survive our cluster-based corrections for statistical significance.

In Author response image 6 we display the non-thresholded version for the reviewers. However, we believe it is not a recommended statistical practice to show non-significant results (after corrections for multiple comparisons), thus we prefer to keep this figure only here. Having said that, we will follow the advice from eLife and the reviewers.

Author response image 6

Download asset Open asset

13) The 9Hz difference in Figure 5B is somewhat unexpected. This is described as a 'stimulus harmonic', but if so, we wouldn't expect to see a difference between matched and mismatched trials. Do the authors have an explanation for this? Does the 9Hz response also correlate with the PSI measure?

We are not surprised by the 9 Hz response, as our data also shows that the 9 Hz response (frequency-double of 4.5 Hz frequency tag) tracks perceptual alternations during binocular rivalry (Author response image 7).

Author response image 7

Download asset Open asset

During mismatched cues, predominantly, a change in percept is experienced, whereby a low frequency percept enters into conscious awareness (as shown in manuscript Figure 2C). These switches result in changes to the frequency-tagged SNR, and as a result, we see an increase in 9 Hz SNR after the onset of a mismatched cue in parieto-occipital regions.

Thus an increase in ITPC at 9 Hz is to be expected for mismatched relative to matched cues, given an increase in SNR at 9 Hz upon changes in visual percepts. Indeed, increases in SSVEP strength have been shown to result from increased ITPC (Kim et al., NatNeuro, 2007).

By contrast, during matched cues, usually no change in percept is experienced. As a result, there is no transient increase in the 9 Hz response.

Neither the 4.5 Hz response or the 9 Hz response correlates with our PSI measures.

14) The model diagrams in Figure 6 are somewhat confusing. For model (a), do the authors mean that the suppressed image emerges back because attention shifts to the mismatched cue first, and thus the mismatched cue enhances the suppressed image? For model (b), it seems surprising that the presence of the green-crossmodal cue would let the attention focus move from the dominant red image to the suppressed red image, before it moves to the dominant green image.

We included this figure to offer a schematic to emphasise the different foci of attention in our suggested interpretations, and not to provide a detailed model of how we think it works. We have clarified our intention in our revised figure legend, emphasizing that this scheme does not imply any detailed mechanisms of how attentional sampling during binocular rivalry should work. Our take-home message was to compare a) sampling between visual and crossmodal cues and b) sampling between dominant and suppressed visual stimuli. We could remove this figure if the editor and reviewers recommend.

15) For correlation scatterplots, it would be helpful to include R and p values in the plot. Are the straight lines in these plots regression lines that predict the y-values using the x-values (i.e. minimizing the error in the y-direction)? Given that both of these are dependent variables with presumably no implied causation, we recommend using Deming regression lines instead, which minimize the absolute error between the line and each data point.

We have now included r and p values in these plots.

The reviewer is correct, the straight lines are least-squares regression lines, predicting the change in PSI based on nITPC, while minimizing error in the y-direction (change in PSI). We do however have an implied causation (predicting the y-values from the x-values), as we are correlating the nITPC strength 0:2 s after stimulus onset to the change in PSI 2:4 s after cue onset.

As an alternative, we present (Author response image 8) the original correlations, with three-regression lines, as per Huang et al. (eLife, 2013).

Given our prior expectation of the implied causality, we would prefer to stick with the linear regression lines, and have amended our figure captions to clarify these.

Author response image 8

Download asset Open asset

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

The three reviewers and I think that you have addressed all the issues raised satisfactorily, but one. The pending issues regards the ITPC calculations and down-sampling vs. up-sampling.

We note that there is a crucial difference in the definition of ITPC "bias": You are talking about minimizing absolute bias (the fact that ITPC is always overestimated when using a finite number of trials, to an extent inversely related to the trial number, e.g. compare Author response image 4D and 4E). Instead, the reviewers are worried about relative bias (the fact that two experimental conditions with equal "true" ITPC may produce very different ITPC values if the number of trials is not equated, e.g. compare Author response image 4A and 4D).

Your upsampling method keeps the absolute bias as low as possible in each condition (given the trial number), but does not reduce relative bias (e.g. compare Author response image 4C and 4D). Only downsampling can remove this relative bias (e.g. compare Author response image 4A and 4E). This comes with the cost of an increase in absolute bias for the high N condition (compare Author response image 4D and 4E), but given that your analysis relies on a comparison of the two conditions, the relative bias must be eradicated; absolute bias changes are just collateral damage.

To summarize:

1) The reviewer made the null assumption that the 2 experimental conditions (matched/mismatched) have the same true ITPC but different numbers of trials. (Here the true ITPC was assumed to be zero, which is the worst-case scenario: no actual phase-locking).

2) When comparing ITPC across the two conditions, a systematic difference was found (Author response image 4A vs. 4D). This difference is necessarily a statistical artifact.

3) The reviewer applied the authors' upsampling method to equate trials; the systematic difference remained (Author response image 4B/4C vs. 4D).

4) The reviewer applied a downsampling method to equate trials; the (artifactual) difference disappeared (Author response image 4A vs. 4E).

5) Conclusion:

The authors are correct that upsampling can preserve the ITPC to have a similar value as that derived from the original dataset (without resampling). But, this is not the point here. The aim of resampling is to let the ITPCs from the two conditions comparable when performing statistical tests. When comparing ITPC from two conditions with unequal trial numbers, upsampling can produce false positives, downsampling will not. Thus, the authors should use downsampling instead of upsampling.

We are pleased to hear that our ITPC resampling procedure is the only remaining issue.

After reviewing your explanation, we are in agreement that reducing the relative bias in ITPC values is preferable in this case.

We have revised our analysis and proceeded by downsampling. Our overall conclusions have not changed, and we have adjusted the statistical values and figures accordingly.

These changes are in the revised manuscript, and include new data for Figures 4 and 5.

All associated source data and figure supplements have also been updated.

We have now addressed all of the reviewers’ concerns and hope our paper is now suitable for publication in eLife.

Author details

1. Matthew J Davidson

   1. School of Psychological Sciences, Faculty of Medicine, Nursing, and Health Sciences, Monash University, Melbourne, Australia
   2. Monash Institute of Cognitive and Clinical Neurosciences, Monash University, Melbourne, Australia

   Contribution

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

   For correspondence

   mjd070@gmail.com

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-2088-040X

2. David Alais

   School of Psychology, The University of Sydney, Camperdown, Australia

   Contribution

   Conceptualization, Resources, Software, Supervision, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-0411-940X

3. Jeroen JA van Boxtel

   1. School of Psychological Sciences, Faculty of Medicine, Nursing, and Health Sciences, Monash University, Melbourne, Australia
   2. Monash Institute of Cognitive and Clinical Neurosciences, Monash University, Melbourne, Australia
   3. School of Psychology, Faculty of Health, University of Canberra, Canberra, Australia

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing—review and editing

   Contributed equally with

   Naotsugu Tsuchiya

   For correspondence

   jeroen.van.boxtel@monash.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2643-0474

4. Naotsugu Tsuchiya

   1. School of Psychological Sciences, Faculty of Medicine, Nursing, and Health Sciences, Monash University, Melbourne, Australia
   2. Monash Institute of Cognitive and Clinical Neurosciences, Monash University, Melbourne, Australia

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing—review and editing

   Contributed equally with

   Jeroen JA van Boxtel

   For correspondence

   Naotsugu.Tsuchiya@monash.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-4216-8701

• 2,609

  Page views

• 378

  Downloads

• 9

  Citations

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