Original ArticlesReduced sensitivity to social priors during action prediction in adults with autism spectrum disorders
Introduction
Diminished social functioning constitutes one of the core features of ASD. In the last decades, an extensive literature has focused on disturbances of mentalizing or Theory of Mind (ToM), i.e. the ability to attribute beliefs and other mental states to oneself and to others, in individuals with ASD (e.g. Baron-Cohen, 1995, Baron-Cohen et al., 1985, Baron-Cohen et al., 1986, Frith, 1989, Happé and Frith, 1996). However, the nature of the impairment affecting action and intention understanding in ASD is still a matter of debate (Hamilton, 2009). While children with autism have impaired or delayed maturation of ToM, adults with high-functioning ASD continue to experience difficulties with understanding others’ intentions in real-life situations, even when they succeed on standard tests for ToM. Questions have also been raised as to whether difficulties reflect selective impairments in mentalizing or disturbances of low-level mechanisms of action perception or of the resonance mechanism supported by the neuron mirror system (Smith and Bryson, 1994, Hughes, 1996, Russell and Jarrold, 1998, Russell and Jarrold, 1999, Zalla, Bouchilloux et al., 2006, Zalla, Labruyere et al., 2006, Cattaneo et al., 2007, Fabbri-Destro et al., 2009, Vanvuchelen et al., 2007, von Hofsten and Rosander, 2012, Zalla et al., 2013).
Using different experimental paradigms, current studies support the notion of preserved goal and action understanding in individuals with ASD (Hamilton, 2009) while difficulties arise when goals and actions are not visible outcomes, and have to be inferred on the basis of the available visual and contextual information (Cattaneo et al., 2007, Gomot and Wicker, 2012, Zalla et al., 2010). These results suggest that difficulties with action understanding in individuals with ASD might arise from impairments in complementing (i.e. contextualizing) the available sensory evidence with other information. Recently, it has been suggested that sensory atypicalities in ASD – such as enhanced sensations, experience of sensory overload, or hypersensitivity – can be explained by a diminished influence of top-down prior expectations on perceptual experience (“hypo-priors”) associated with increased reliance on sensory evidence, possibly as a consequence of enhanced “bottom-up” functioning (Lawson et al., 2014, Van de Cruys et al., 2014; see also Pellicano & Burr, 2012). Such abnormal interplay between top-down priors and bottom-up sensory evidence may provide a simple explanation for ASD difficulties in inferring other people’s intentions, by yielding significant deviations from normative Bayesian inference (McKay, 2012).
Bayesian models of intention understanding posit that our brain is constantly engaged in the process of drawing inference on the basis of two distinctive types of information: the sensory evidence conveyed by movement kinematics and the observer’s prior expectations about which intention is the most likely cause of what is observed, given past experience (Baker et al., 2009, Baker et al., 2006, Chambon, Domenech et al., 2011, Kilner, 2011). Intention inference is contingent upon an adaptive interplay between these two sources of information, with observers tending to rely progressively more on their prior expectations as the reliability of sensory evidence decreases, and vice versa. Crucially, this interaction has also been found to vary according to the ‘type’ of intention to be inferred, with participant’s prior experience gaining weight over sensory evidence when inferring motor intentions directed at a third party (Social intentions) rather than motor intentions directed at objects in isolation (Non-Social intentions) (Chambon, Domenech et al., 2011, Chambon, Pacherie et al., 2011).
Here, in accordance with theoretical models (Lawson et al., 2014, Pellicano and Burr, 2012), we hypothesized that diminished mentalizing abilities in individuals with ASD could be accounted for by an abnormal weighting of these two types of information (prior expectations and sensory evidence), in turn depending on the type of intention to be inferred (Social vs. Non-Social intentions). Specifically, attenuated “top-down” priors might be responsible for difficulties encountered by ASD in inferring social intentions. Indeed, social intentions often show a complex relationship with the behaviour they cause, i.e., in social situations many competing intentions are potentially consistent with what is observed. As such, social intentions cannot be unambiguously inferred from mere observation, and prior knowledge is required to constrain the space of candidate intentions (Chambon, Domenech et al., 2011, Kilner et al., 2007a, Kilner et al., 2007b). If individuals with ASD do show attenuated priors, one would expect deficits to be more severe in social situations where priors are massively required (Lawson et al., 2014). Thus, particularly relevant to the autistic symptomatology is the hypothesis that poor “social-specific” priors may result in an incapability to reduce the intrinsic uncertainty of social behaviours, in turn compromising one’s ability to effectively interact with others (Chambon, Pacherie et al., 2011, Sinha et al., 2014; see also Fletcher & Frith, 2009). Although not yet supported by direct empirical evidence, this hypothesis is otherwise consistent with observations made in individuals with schizophrenia suffering from negative (“autistic-like”) symptoms. While these patients exhibit poor prior expectations in the social domain, they also rely more strongly on sensory evidence to infer intentions when those are directed at a third party rather than at a non-meaningful object (Chambon, Pacherie et al., 2011). Interestingly, it has been suggested that, in these patients, social situations may not prompt the same expectations as those typically observed in comparison participants, leading to the formation of abnormal (unreliable) predictions about others’ social intentions (Zalla et al., 2004, Zalla, Bouchilloux et al., 2006, Zalla, Labruyere et al., 2006, Barbalat et al., 2009, Barbalat et al., 2011, Chambon, Pacherie et al., 2011, Chambon et al., 2012).
In the present study, we aimed to test this hypothesis by contrasting the ability to infer non-social and social intentions in a group of adults with ASD, as compared to a control group of typically developed adults. Participants were presented with a series of video clips showing an actor manipulating (either transporting or rotating) a non-meaningful object and asked to infer intentions directed at an object or intentions directed at a third party, respectively. In both the “Social” and the “Non-Social” intention tasks, the specific contributions of sensory evidence and prior expectations to the intentional inference were systematically manipulated by varying the amount of visuo-motor evidence conveyed by the action scene (i.e. the completeness of action sequences) and the probability of occurrence associated with each different intention. We then looked at (i) whether participants’ performance on each condition could be accounted for by an abnormal dependence on visuo-motor evidence and/or prior expectations, depending on the type (Social vs. Non-Social) of the intention being manipulated, and (ii) whether this abnormal dependence – if observed – correlated with the severity of autistic symptoms, as measured by the Autism Diagnostic Interview–Revised (ADI-R) (Lord, Rutter, & Le Couteur, 1994).
Section snippets
Participants
Eighteen adult participants with ASD and twenty comparison participants (CP) were recruited to participate in the study from Albert Chenevier Hospital in Créteil. The groups were matched for age, education, gender and full-scale IQ, as measured by the Wechsler Adult Intelligence Scale (Wechsler, 1999) (see Table 1). All participants were screened for exclusion criteria (dyslexia, epilepsy, and any other neurological or psychiatric conditions) prior to taking part to the study. Participants in
Discriminability index (d′) and response bias (c)
Two types of analyses were conducted. Correct Response rates (CR) were first analysed within each group using a 2 × 2 × 4 × 2 repeated-measures ANOVA with Group (Comparison versus ASD) as a between-subjects factor, and type of Intention (Non-Social versus Social), Amount of visual information (low, moderate, high, and very high) and Prior (likely versus unlikely intentions), as within-subjects factors (see Fig. 2, and Supplementary analyses).
Then, we aimed to quantify independently the effect of
Discussion
The aim of the present study was to investigate whether in adults with ASD poor performance in inferring intentions might be explained by an unbalanced interplay between top-down prior expectations and bottom-up sensory stimulation. Specifically, we expected to find reduced reliance on prior expectations when it comes to inferring social intentions in participants with ASD. We further hypothesized that such reduced dependence, if observed, would correlate with the severity of autistic symptoms
Acknowledgements
We gratefully acknowledge the commitment of the participants and their families to the pursuit of research in autism. This research was supported by an ANR Grant (ANR-ZR81) to CF, VC, EP and TZ, and by Fondation FondaMental and Fondation de France (00041930) to TZ and ML. VC was supported by a postdoctoral fellowship of the Région Ile-de-France (Paris) and a research grant from the Fyssen Foundation. VC, EP and TZ were supported by ANR-10-LABX-0087 IEC and ANR-10-IDEX-0001-02 PSL∗. Designed
References (63)
- et al.
Perfect reciprocity is the only evolutionarily stable strategy in the continuous iterated prisoner's dilemma
Journal of Theoretical Biology
(2007) - et al.
Action understanding as inverse planning
Cognition
(2009) - et al.
Impaired hierarchical control within the lateral prefrontal cortex in schizophrenia
Biological Psychiatry
(2011) - et al.
Does the autistic child have a ‘theory of mind’?
Cognition
(1985) Alternative Bayesian accounts of autistic perception: Comment on Pellicano and Burr
Trends in Cognitive Sciences
(2012)- et al.
Autism spectrum disorders: Developmental disconnection syndromes
Current Opinion in Neurobiology
(2007) - et al.
A challenging, unpredictable world for people with autism spectrum disorder
International Journal of Psychophysiology
(2012) - et al.
Predictive coding explains binocular rivalry: An epistemological review
Cognition
(2008) More than one pathway to action understanding
Trends in Cognitive Sciences
(2011)- et al.
The felt presence of other minds: Predictive processing, counterfactual predictions, and mentalising in autism
Consciousness & Cognition
(2015)