Modality-specific and modality-independent components of the human imagery system

Abstract

Imagery research typically deals with the commonalities and differences between imagery and perception. As such, it is usually confined to one specific modality. Yet, it is likely that some of the underlying processes are shared between different sensory modalities while others are modality-specific. In this fMRI study, we used a balanced design that allowed for a direct comparison between imagery and perception in visual and auditory modalities, and also for a link between subjective imagery experience and brain activation. Results indicated a selective role for the “default mode network” as a modality-independent “core” imagery network. In addition, results identified areas in the visual and auditory association cortices that contributed to mental imagery in a modality-specific fashion. Interestingly during mental imagery, primary visual and auditory cortices showed modality-specific suppression of activity. This is the first fMRI study to characterize both modality-specific and modality-independent components of the human imagery system.

Introduction

Functional neuroimaging studies have identified a consistent set of brain structures supporting our ability to perceive with the mind's eye and ear — also referred to as mental imagery. These regions generally overlap with the same regions that are involved in “real-world” visual and auditory perception. Yet, neuroimaging research of mental imagery has been confined almost exclusively within the boundaries of a single sensory modality, either visual or auditory, and has focused specifically on areas within the perceptual processing system of that particular modality. Moreover, the single neuroimaging study that did compare visual and auditory imagery modalities within the same experiment did not employ comparable task conditions (Halpern et al., 2004). Thus, it remains unclear whether there might be higher-level cognitive processes that are mediated by a network outside of the sensory cortices supporting all types of mental imagery. To our knowledge, the existence of such a modality-independent “core” imagery network in the brain has never been tested.

In the present fMRI study, we addressed this issue by focusing on brain regions involved in both auditory and visual imagery under matching task conditions. Inside the MRI scanner, participants either saw pictures or heard sounds (perceptual conditions) or actively imagined pictures or sounds (imagery conditions) associated with specific cue words over a 3-second period. Immediately following each imagery trial, participants rated the subjective quality of the imagery experience (Fig. 1). In order to address an important confounder, we later asked participants outside of the scanner to rate how well they were able to separate the visual and auditory imagery modalities for the words previously presented inside the scanner. This design allowed us to investigate, under matching task conditions, not only modality-specific, but also modality-independent, components of the human imagery system.

Potential brain areas that might support a modality-independent role in mental imagery involve those implicated in the “default mode network” (DMN). The DMN is a network of brain regions that tends to be more active during passive rest conditions than during the performance of demanding cognitive tasks, and includes posterior cingulate, lateral posterior parietal, and medial prefrontal regions (Buckner et al., 2008, Raichle et al., 2001). Recent studies have further indicated an active role of the DMN in internally-oriented cognitive processes such as mind wandering (Mason et al., 2007), self-referential mental activity (Gusnard et al., 2001), thinking about the past (Daselaar et al., 2006, Hassabis et al., 2007a, Huijbers et al., 2009), and envisioning the future (Szpunar et al., 2007). Importantly, a recent study also linked the DMN to visual imagery and suggested that the network might subserve a more general role in multi-modal constructive processes supporting mental imagery and other internal operations (Hassabis et al., 2007a). This idea was directly tested in the present study by including both auditory and visual imagery modalities, which allows examining not only differences but also commonalities between these modalities.

Regarding regions supporting modality-specific components of imagery, considerable evidence has been collected implicating areas that are generally associated with perception. Several studies have indicated considerable overlap between visual perception and visual imagery (Chen et al., 1998, Ishai et al., 2000, Kosslyn et al., 1999, O'Craven and Kanwisher, 2000). Yet, the extent of overlap remains unclear (Kosslyn et al., 2001, Kosslyn and Thompson, 2003a). Whereas most studies found overlap only in visual association (VA) cortices, including V2, V3, and V4 (Amedi et al., 2005, Ishai et al., 2000, Knauff et al., 2000), some studies also found imagery-related activity in primary visual (V1) cortex (Chen et al., 1998, Kosslyn et al., 1999, Slotnick et al., 2005). Likewise, considerable overlap has been found between regions supporting auditory perception and imagery (Bunzeck et al., 2005, Halpern and Zatorre, 1999, Yoo et al., 2001, Zatorre and Halpern, 2005), but again the extent of overlap within primary (A1) and association auditory (AA) cortices remains unclear. By directly comparing visual and auditory imagery under matching task conditions, the present study can shed more light on the role of primary and secondary sensory regions in mental imagery.

The present study tested two main predictions. First, based on its presumed role in internal processing, we predicted that the DMN regions would form a “core” imagery network and show modality-independent imagery-related activity. In other words, we expected that both auditory and visual imagery would yield more DMN activity than auditory and visual perception. We also predicted that these modality-independent activations would be coupled with subjective imagery experience as assessed by the imagery scale. Second, given that most imagery studies did not find overlap with perception in primary sensory regions, we predicted that only AA and VA would show modality-specific imagery-related activity. In other words, we expected more activity in VA during visual imagery vs. auditory imagery as well as during visual perception vs. auditory perception. Similarly, we predicted more activity in AA during auditory imagery (A-IMG) vs. visual imagery (V-IMG) as well as during auditory perception (A-PCP) vs. visual perception (V-PCP). Similar to the modality-independent regions, we also predicted that the modality-specific activations would be associated with subjective imagery experience as assessed by the imagery quality ratings for the corresponding modalities. In other words, we predicted visual imagery regions to correlate with the visual, but not auditory, imagery ratings, and auditory imagery regions with the auditory, but not visual, imagery ratings.