Attentional orienting triggered by gaze in schizophrenia

Abstract

The ability to detect the direction of another person's gaze and to shift our own attention reflexively in the same direction facilitates the sharing of attention with other people. Such sharing of attention would seem critical for the maintenance of normal social cognition. Social cognition is severely impaired in people with schizophrenia. So, we used spatial cuing paradigms to investigate reflexive (Experiment 1) and controlled (Experiment 2) attentional orienting triggered by gaze in schizophrenia. In Experiment 1, 30 patients and 24 controls detected targets appearing right or left of a central image of a head turned right, left, or straight-ahead. These gaze-cues were non-predictive. Patients, but not controls, showed a significant congruency advantage at 100 ms SOA. The congruency advantage was similar in patients and controls at 300–800 ms SOA. In Experiment 2, 20 patients and 24 controls detected targets 300–800 ms after a central gaze-cue that pointed away from the target on 80% of trials. Controls, but not patients, were able to reverse the reflexive congruency advantage at 800 ms SOA. This study provides the first evidence that people with schizophrenia show abnormally sensitive gaze-triggered reflexive orienting. Findings are discussed in light of recent neuroimaging work investigating the neural basis of social orienting and social cognition.

Introduction

Human beings are highly evolved social creatures with a keen awareness of themselves and other people as co-experiencing subjective beings. A full appreciation of the inter-subjective nature of social life depends upon a domain of mental processing termed social cognition. This domain depends upon the more basic social perception skills (e.g. facial affect recognition) as well as higher order “theory-of-mind” abilities; theory-of-mind refers to a capacity to impute causal mental states in order to explain and/or predict behavior in terms of psychological causation (Premack & Woodruff, 1978).

There is mounting evidence from neuropsychological and neuroimaging studies to support the view that human beings have evolved a dedicated neural network to sustain social cognition. For example, individuals with Williams’ syndrome display relatively spared social cognition skills alongside impairments in other domains (e.g. spatial cognition), while individuals with autism display pervasive social cognition deficits alongside spared capacities in other domains (e.g. spatial cognition and fluid intelligence; see Pinkham, Penn, Perkins, & Lieberman, 2003 for discussion). Findings of this type provide evidence of a putative double dissociation between social cognition and other cognitive domains and support the existence of a functionally independent neural network for social cognition.

Neuroimaging studies of social cognition, in particular those that have investigated theory-of-mind, consistently identify three key regions as critical for social cognition: the superior temporal sulcus (STS), the medial prefrontal cortex (MPFC) and the temporal poles (see Frith & Frith, 2003; Gallagher & Frith, 2003 for recent reviews). The role of the STS in this neural network is to detect “intentionality” (i.e. an agent's mental directedness towards the environment), which is signaled primarily by the agent's direction of gaze and also by the agent's purposive biological motion. The role of the MPFC is to link intentional agents to their environment via mental states that are about the world, yet separate from the world, and contingent upon a unique subjective viewpoint. Finally, the temporal poles may contribute to inferring the specific content of another person's mental state by retrieving memories of past events that resemble the other person's circumstances. Researchers who broaden the construct of social cognition to also include social cue perception (e.g. facial affect recognition), the detection of socially salient stimuli, reactions to violations of social norms, and reasoning about normative social behaviour also highlight the role of the fusiform gyrus, amygdala, and orbitofrontal cortex (see Adolphs, 1999, Adolphs, 2001; Siegel & Varley, 2002 for reviews).

In some sense, the STS might be considered the linchpin of the social cognition network. Neurophysiological and neuroimaging studies have provided evidence that the STS is responsible for the determination of another person's direction of gaze and the triggering of reflexive shifts of attention in the same direction. For example, Perrett et al., 1985, Perrett et al., 1987, Perrett et al., 1992 have identified specific cells in the STS that respond to gaze-direction in non-human primates. Neuroimaging studies of gaze-detection in humans have also implicated the STS; the finding of concurrent activation in the intraparietal sulcus suggesting that the determination of gaze-direction automatically recruits the spatial cognitive system in order to shift attention reflexively in the same direction as another person's gaze (Hoffman & Haxby, 2000). In a recent review, Emery (2000) highlighted the amygdala and the orbitofrontal cortex, as well as the STS, as playing key roles in the processing of gaze-direction information. According to Emery, the STS most likely codes the direction of another person's gaze, as well as the focus of their attention, whereas the amygdala may attach “emotional significance to the eyes” (Emery, 2000, p. 597), with recent evidence suggesting that heightened amygdala activity may facilitate the processing of gaze-direction cues in healthy adults (Hooker et al., 2003). Finally, the orbitofrontal cortex may be responsible for evaluating the social significance of another person's direction of gaze.

Why should rapid detection of another person's direction of gaze and reflexive shifts of attention in the same direction (subserved by the STS) be the linchpin of social cognition? We suggest this is because the eyes are a unique facial feature; they are the “windows to the soul” (Baron-Cohen, 1995). According to Baron-Cohen, young children first experience a basic “meeting of minds” with other people by attending to other people's eyes and shifting their own attention in response to another person's gaze-direction in order to share attention with that other person. Such sharing of attention entails a mutual awareness that self and other see the same thing (i.e. share a similar subjective experience which is directed towards the same object), thus providing a crucial foundation for the development of normal social cognition in the young child. However, sharing attention with other people may be critical, not only for the development of normal social cognition, but also for the maintenance of normal social cognition; in other words, we may stay sensitized to the subjective lives of other people by regularly sharing attention with them. The ability to rapidly detect other people's gaze-direction and to shift attention reflexively in the same direction enables us to engage spontaneously and effortlessly in sharing attention with other people. According to this account, people with abnormal STS function, who are impaired in their ability to automatically detect gaze-direction and to shift attention reflexively, will engage less in sharing attention, thus compromising their social cognition.

Schizophrenia has been characterized as a social disorder, and performance deficits on social cognition tasks are prominent in schizophrenia (for reviews, see Corcoran, 2001; Langdon et al., 2001a, Langdon et al., 2001b, Langdon et al., 2002; Penn, Corrigan, Bentall, Racenstein, & Newman, 1997). This paper examines whether social deficits in schizophrenia reflect a more basic disturbance of automatic gaze-detection and reflexive orienting triggered by gaze. Consistent with this hypothesis, the same regions that play a role in the neural networks for social orienting and social cognition have been implicated in the neuropathy of schizophrenia (e.g. the amygdala, MPFC and orbitofrontal cortex; see Emery, 2000, Grossberg, 2000; Gur & Pearlson, 1993; Pinkham et al., 2003 for reviews). In particular, neuroimaging (Gur & Pearlson, 1993) and electrophysiological (Levitan, Ward, & Catts, 1999; McCarley et al., 1993) studies have provided evidence that anomalies of the STS are present in some people with schizophrenia (e.g. P300 abnormalities and hallucinatory experiences are more severe in schizophrenia patients with reduced superior temporal gyral volumes). Furthermore, visual-scanpath studies have found that people with schizophrenia spend less time looking at other people's facial features, in particular, their eyes (Green, Williams, & Hemsley, 2000; Loughland, Williams, & Gordon, 2002). Finally, Rosse et al. (1994) have found that people with schizophrenia make errors when asked to report the direction of another person's gaze.

In the Rosse et al. study, however, participants were shown slides of faces and asked explicitly “is the person in the slide looking directly at you?” That people with schizophrenia show a difficulty with making conscious judgments of gaze-direction need not mean that these individuals are impaired in their ability to automatically detect another person's gaze and to shift attention reflexively in the same direction. The aim of Experiment 1 is to investigate reflexive orienting triggered by another person's gaze-direction in people with schizophrenia.

The traditional paradigm used to investigate attentional orienting is the spatial cuing task (Posner, 1980). In a typical cuing task, participants are asked to press a button as soon as they see a target; the target can appear on one side or other of a computer screen, and the participant's attention will have just been cued either in the correct direction (a “congruent” cue) or in the incorrect direction (an “incongruent” cue).¹ Reaction times (RTs) are reliably faster when targets follow congruent cues, although the time-course of congruency effects differs depending upon the type of directional cue (see below).

Mainstream cuing studies of reflexive versus controlled orienting make a distinction between exogenous orienting, which is conceived as stimulus-driven and reflexive, and endogenous orienting, which is conceived as consciously mediated and amenable to voluntary control (see Driver et al., 1999 for a review). Langton and Bruce (1999) have proposed that social attention signals (e.g. the direction of another person's gaze) may engage exogenous orienting mechanisms. Exogenous orienting typically occurs when a peripheral event, such as a brief flash of light appears in the right or left visual field; the orienting which occurs in such circumstances is fast (participants are faster detecting targets 100 ms after a congruent cue), short-lived (differences between congruent and incongruent RTs disappear by 700–800 ms if a cue is non-predictive) and obligatory (at short cue-target intervals, participants cannot stop shifting their attention in the direction of a brief flash of light even when that brief flash of light appears on the non-target side 80% of the time).

In contrast, endogenous orienting occurs in response to symbolic directional cues (e.g. the word “RIGHT” or “LEFT” positioned centrally, or central arrows) when these are predictive of target location. Here, cues need to be interpreted semantically in order to derive directional information before they can generate an orienting response. Endogenous orienting is more slowly generated (maximal facilitation requires cue-target intervals of 300–400 ms; Cheal & Lyon, 1991; Müller & Rabbitt, 1989), persists longer (so long as observers expect cues to be informative), and can occur even when cues are incongruent, given that cues are predictive and observers have sufficient time to exert voluntary control (e.g. when a central arrow points away from the target 80% of the time and cue-target intervals extend to 700 ms or more, healthy observers can endogenously redirect their attention such that incongruent RTs become faster than congruent RTs).

Friesen and Kingstone (1998) and Langton and Bruce (1997) were the first to adapt spatial cuing paradigms in order to investigate attentional orienting triggered by gaze. Friesen and Kingstone (1998) used schematic line drawings of faces (positioned centrally) with the eyes shifted right or left. Although their gaze-cues were non-predictive, they found a significant congruency advantage (whether to detect, localize or identify a target) at cue-target intervals as short as 105 ms. Langton and Bruce, 1997, Langton and Bruce, 1999 used realistic photographs of faces, rather than schematic line drawings. They cued target location using a central image of a head turned right, left, up, or down and asked participants to detect targets as quickly as possible. Langton and Bruce also found facilitation for congruent orientation of head and eyes, despite their cues being non-predictive. The facilitatory effects found by Langton and Bruce were maximal at 100 ms stimulus onset asynchrony (SOA), small but non-significant at 500 ms SOA, and non-existent at 1000 ms SOA. In the same year, however, Driver et al. (1999) cued target location using realistic photographs of faces with the eyes pointing right or left (i.e. the head did not turn, similar to the schematic line drawings used by Friesen and Kingstone). They found that it took 300 ms before non-predictive congruent gaze-cues triggered reliable facilitatory effects, much longer than the 100 ms SOA found by Langton and Bruce and by Friesen and Kingstone to be sufficient to induce significant facilitatory effects, also much longer than the 100 ms SOA considered sufficient to induce significant facilitatory effects from non-predictive peripheral cues. Furthermore, the congruency advantage found by Driver et al. persisted until at least 700 ms SOA. This was unlike the findings of Langton and Bruce who reported non-significant congruency effects at 500 ms SOA. It was also unlike the typical time-course of congruency effects triggered by non-predictive peripheral spatial cues. These cues are considered the classic trigger of exogenous orienting and induce a rapid congruency advantage which quickly diminishes and then reverses at around 200–250 ms SOA (i.e. incongruent RTs become faster than congruent RTs). This reversal effect is known as inhibition of return (IOR; see Klein and MacInnes, 1999 for further discussion).

Nevertheless, despite the differences between the Driver et al. (1999) findings and the classic findings of peripheral spatial cuing studies, Driver et al. concluded that the facilitatory effects triggered by their congruent gaze-cues at 300 ms SOA were indicative of reflexive orienting. This was because the gaze-cues were non-predictive and because participants in a later experiment (also reported in Driver et al., 1999) were unable to consciously inhibit orienting attention in the direction of another person's gaze at 300 ms SOA, even though the targets were now four times more likely to appear in uncued locations.

More recently, Downing et al. (2004) and Langdon and Smith (in press) have found that a 100 ms SOA may be too short for realistic gaze-cues, of the type used by Langton and Bruce and by Driver et al. to reliably trigger reflexive shifts of attention in healthy adults, and that SOAs of at least 200–300 ms duration are needed to induce such effects. Langdon and Smith also found that realistic depictions of gaze, unlike central arrows, can trigger a long-lasting congruency advantage (e.g. up to at least 800 ms SOA), even when the cues are non-predictive, so long as the gaze-cues remain visible and participants maintain their preparedness to respond. This may help to explain why Langton and Bruce (1999) and Driver et al. (1999) found a different time-course of reflexive orienting triggered by gaze, i.e. cue and target remained together on screen until response in the Driver et al. study, whereas the cue disappeared when the target appeared in the Langton and Bruce study.

In sum, while realistic cues of gaze-direction resemble symbolic directional cues (e.g. central arrows) in being slower than peripheral spatial cues to trigger attentional shifts, gaze-cues also appear to resemble peripheral spatial cues in triggering reflexive shifts of attention that are impervious to conscious control at relatively short SOAs (i.e. in the order of 300 ms). The aim of Experiment 1 is to investigate reflexive orienting triggered by realistic cues of gaze-direction in people with schizophrenia. Evidence of a lack in schizophrenia of the normal orienting response reflexively triggered by another person's direction of gaze would help to support the view that the STS plays a critical linchpin role in the neural networks for social orienting and social cognition, and that anomalies of the STS contribute to social deficits in schizophrenia.