Interregional alpha-band synchrony supports temporal cross-modal integration

Highlights

• We investigated EEG dynamics of cross-sensory integration during time perception

• Psychometric functions were fitted on behavior using Bayesian graphical modeling

• There was strong auditory dominance when subjects judged audiovisual duration

• This correlated with audiovisual interregional alpha (8-12 Hz) phase synchrony

• Single-trial alpha phase synchrony predicted the degree of psychometric lapsing

Abstract

In a continuously changing environment, time is a key property that tells us whether information from the different senses belongs together. Yet, little is known about how the brain integrates temporal information across sensory modalities. Using high-density EEG combined with a novel psychometric timing task in which human subjects evaluated durations of audiovisual stimuli, we show that the strength of alpha-band (8–12 Hz) phase synchrony between localizer-defined auditory and visual regions depended on cross-modal attention: during encoding of a constant 500 ms standard interval, audiovisual alpha synchrony decreased when subjects attended audition while ignoring vision, compared to when they attended both modalities. In addition, alpha connectivity during a variable target interval predicted the degree to which auditory stimulus duration biased time estimation while attending vision. This cross-modal interference effect was estimated using a hierarchical Bayesian model of a psychometric function that also provided an estimate of each individual's tendency to exhibit attention lapses. This lapse rate, in turn, was predicted by single-trial estimates of the stability of interregional alpha synchrony: when attending to both modalities, trials with greater stability in patterns of connectivity were characterized by reduced contamination by lapses. Together, these results provide new insights into a functional role of the coupling of alpha phase dynamics between sensory cortices in integrating cross-modal information over time.

Introduction

When the sound coming from a television gets out-of-sync with what is visually displayed, you immediately feel that something is wrong. Multimodal processing is ubiquitous in perception: different senses provide us with complementary evidence about external events, which can aid our responses to these events (McDonald et al., 2000, Yang et al., 2013), or can result in perceptual illusions (Alais and Burr, 2004, McGurk and MacDonald, 1976). Over the past several decades, neuroscience of multisensory processing has shifted from a strict hierarchical view of unisensory signals converging onto higher supramodal areas (Meredith and Stein, 1983, Stein and Stanford, 2008), to a growing consensus that cross-modal integration can occur even at early stages of sensory processing (Foxe et al., 2000, Ghazanfar and Chandrasekaran, 2007, Ghazanfar and Schroeder, 2006, Giard and Peronnet, 1999, Kayser and Logothetis, 2007, Martuzzi et al., 2007, Molholm et al., 2002). However, how these early-stage interactive processes are neurophysiologically organized remains a topic of active exploration (Klemen and Chambers, 2012, Sarko et al., 2013).

One proposed mechanism of neural interaction is “binding through coherence” (Fries, 2005, Varela et al., 2001, Ward, 2003, Womelsdorf et al., 2007), or the idea that effective windows of cortico-cortical communication may arise by transiently synchronized electrophysiological oscillations of the involved neural populations. Evidence is accumulating that this principle may apply to the integration of multisensory information as well (Doesburg et al., 2008, Hummel and Gerloff, 2005, Sarko et al., 2013, Senkowski et al., 2008, von Stein et al., 1999). For example, a stimulus of one modality can modulate the processing of a concurrently delivered stimulus of another modality, through the phase resetting of ongoing oscillatory activity in the corresponding primary sensory region (Diederich et al., 2012, Kayser et al., 2008, Lakatos et al., 2007). This results in increased cortical excitability, and thus improved behavioral performance towards the bimodal stimulus. Attention seems to determine which modality “controls” the phase-resetting (Lakatos et al., 2009). Given the tight link between alpha-band (8–12 Hz) activity and attentional processing (Jensen and Mazaheri, 2010, Klimesch et al., 2007), we hypothesized that during cross-modal attention, alpha phase synchrony may be an important mediator of large-scale communication between distant sensory regions (Hummel and Gerloff, 2005, Palva and Palva, 2011). Although several frequency bands have been implicated in multisensory integration (Senkowski et al., 2008), it has been proposed that phase synchronization in the alpha-band may be especially important for coordinating functional integration between cortical regions (Doesburg et al., 2009) of relatively longer inter-areal distances (Palva and Palva, 2007, Palva et al., 2005). This active role of alpha activity has been shown in a variety of cognitive and perceptual tasks, such as spatial attention (Doesburg et al., 2009), working memory (Palva et al., 2005), object recognition (Bar et al., 2006), and error-processing (van Driel et al., 2012), and may thus represent a general mechanism of coherent network functioning. Importantly, increases in interregional alpha-band phase synchrony can co-occur with local decreases in alpha-band power (Palva and Palva, 2007, Palva and Palva, 2011), where the latter may reflect attention-related “active inhibition” of task-irrelevant areas (Jensen and Mazaheri, 2010). In this study, we were in particular interested in the role of interregional alpha phase dynamics during cross-sensory integration.

Most studies on multisensory processing use brief, momentary stimuli, and focus on the spatial domain (Driver and Spence, 1998, Driver and Spence, 2000, Macaluso and Driver, 2005), or on judgments of successiveness versus simultaneity (Jaekl and Harris, 2007, Keetels and Vroomen, 2007). However, multisensory events in a continuous environment are more likely to be prolonged (Ghazanfar and Chandrasekaran, 2007), and are not necessarily linked to one spatial location; in these more naturalistic situations, correlated temporal durations can provide key evidence for integration. Moreover, it is especially interesting to study cross-modal integration of elapsed time, because the perception of auditory duration is superior to that of visual duration. This is in contrast to the more frequently investigated spatial domain, in which visual spatial localization is superior to auditory spatial localization (Burr et al., 2009, Fendrich and Corballis, 2001, Pick et al., 1969).

The purpose of the present study was to investigate the potential neural mechanisms of multimodal integration via duration perception. We here report novel evidence that inter-regional phase synchrony in the alpha-band supports multimodal duration judgments in humans. Through time–frequency decomposition of high-density EEG activity, we found that alpha synchrony was modulated by cross-modal attention and correlated with subject-specific Bayesian estimated parameters of distractor interference and lapsing.