When seeing depends on knowing: Adults with Autism Spectrum Conditions show diminished top-down processes in the visual perception of degraded faces but not degraded objects

Abstract

Behavioural, neuroimaging and neurophysiological approaches emphasise the active and constructive nature of visual perception, determined not solely by the environmental input, but modulated top-down by prior knowledge. For example, degraded images, which at first appear as meaningless ‘blobs’, can easily be recognized as, say, a face, after having seen the same image un-degraded. This conscious perception of the fragmented stimuli relies on top-down priming influences from systems involved in attention and mental imagery on the processing of stimulus attributes, and feature-binding [Dolan, R. J., Fink, G. R., Rolls, E., Booth, M., Holmes, A., Frackowiak, R. S. J., et al. (1997). How the brain learns to see objects and faces in an impoverished context. Nature, 389, 596–599]. In Autism Spectrum Conditions (ASC), face processing abnormalities are well-established, but top-down anomalies in various domains have also been shown. Thus, we tested two alternative hypotheses: (i) that people with ASC show overall reduced top-down modulation in visual perception, or (ii) that top-down anomalies affect specifically the perception of faces. Participants were presented with sets of three consecutive images: degraded images (of faces or objects), corresponding or non-corresponding grey-scale photographs, and the same degraded images again. In a passive viewing sequence we compared gaze times (an index of focal attention) on faces/objects vs. background before and after viewers had seen the undegraded photographs. In an active viewing sequence, we compared how many faces/objects were identified pre- and post-exposure. Behavioural and gaze tracking data showed significantly reduced effects of prior knowledge on the conscious perception of degraded faces, but not objects in the ASC group. Implications for future work on the underlying mechanisms, at the cognitive and neurofunctional levels, are discussed.

Introduction

Autism Spectrum Conditions (ASC, including autism, atypical autism, and Asperger Syndrome) are severe neuro-developmental conditions characterised by the co-occurrence of impairments in social interaction, communication, and restricted and repetitive behaviours. Despite considerable progress, the nature of the underlying abnormalities at the cognitive and neurobiological levels, is not fully understood.

One line of research suggests that deficits in social cognition and in an underlying ‘social brain network’ represent the core problem (Baron-Cohen, 2003, Baron-Cohen et al., 2000, Frith and Frith, 1999, Hobson, 2002, Klin et al., 2003, Mundy and Neal, 2001). Since many aspects of successful social interaction involve processing information from faces, over recent years abnormalities in face processing have become a key area of research in ASC (Dawson et al., 2005, Schultz, 2005). Behavioural studies have shown deficits and/or abnormalities in different aspects of face processing, including the ability to recognize social communicative cues (emotion expressions, eye gaze) from faces (Golan et al., 2006, Hobson, 1986, Katsyrie et al., 2008, Teunisse and de Gelder, 2003) and the ability to identify/remember individual face identities (Boucher and Lewis, 1992, Klin et al., 1999, Langdell, 1978, Joseph and Tanaka, 2003), though several negative findings also exist (see Jemel, Mottron, & Dawon, 2006, for a review).

Early neuroimaging studies found reduced activation in the lateral fusiform gyrus (a region also called “fusiform face area” [FFA] because it typically responds more strongly to faces than any other object class; Kanwisher, McDermott, & Chun, 1997) when viewers with ASC looked at faces, suggesting that abnormalities in the development or function of this brain region may be critical to the face processing abnormalities (Critchley et al., 2000, Pierce et al., 2001, Schultz et al., 2000). However, more recent studies showed relatively normal FFA activation in ASC (Bird et al., 2006, Dapretto et al., 2006, Hadjikhani et al., 2004, Hadjikhani et al., 2007) but a lack of the modulation of FFA activity by attention (fronto-parietal network, Bird et al., 2006) or emotional significance (notably the amygdala, Kleinhans et al., 2008, Pierce and Redcay, 2008), as is seen in typically developing volunteers (Vuilleumier et al., 2001). These findings are consistent with the proposal that diminished spontaneous attention to faces, perhaps because they do not convey the same significance or reward value as for other people, contributes to the face processing abnormalities in ASC (Klin et al., 2003).

A second approach to understanding ASC focuses on domain-general perceptual and cognitive abnormalities that may, in principle, play a role in the social impairments of ASC, but also in their non-social features, including exceptional perceptual abilities. In particular, a great deal of research has been inspired by the notion of weak central coherence, which refers to a cognitive style in which the normal tendency to integrate information in context and for meaning is diminished, with preferential focus on details and local information (Frith, 1989, Happé and Frith, 2006). In support of this, several studies have shown that people with ASC outperform comparison groups on tests where local processing and/or resistance to the overall gestalt, are advantageous; for example, the Embedded Figures and Block Design Tasks, and tests of visual illusions (Happé, 1996, Happé and Booth, 2008) and visual search (Plaisted, Swettenham, & Rees, 1999, but see, for example, Mottron et al., 1999, Ropar and Mitchell, 2001, for negative or conflicting findings).

Over recent years, the weak coherence hypothesis has evolved in two important directions: Some theorists have emphasised enhanced local processing abilities, distinct from global/contextual impairments (Mottron and Burack, 2001, Mottron et al., 2006, Plaisted, 2001). The study reported here is part of a second trend, which has revisited the notion of difficulties in ‘integrating information in context’. Earlier studies primarily focused on contextual processing in a Gestaltist sense, which requires the integration of different visual elements to form a coherent or meaningful whole, such as in the perception of Kanizka triangles (see, for example, Happé & Frith, 2006, for a review). However, “context” effects in visual perception also include the influence of one's past experience and prior knowledge of the world on the perception of an object, independent of or in addition to the intrinsic properties of the object itself. Indeed, over recent years there has been an increasing interest in abnormalities in the influence of prior knowledge in perception and cognition in ASC. (C. Frith, 2003, Frith and Frith, 2006, Greenaway and Plaisted, 2005, Loth, 2007, Loth et al., 2008, Ropar and Mitchell, 2002). This relates to processes commonly referred to as top-down processes, and they are reviewed next.