While impaired pragmatics constitutes one of the most robust characteristics of autism, independent of linguistic or developmental level, Jaswal & Akhtar devote it virtually no space within their criticisms of the social motivation theory of autism. In this commentary, I complement their discussion by arguing that data on pragmatics in autism, especially on the comprehension side, speak against such motivation-based accounts.

Current diagnostic criteria for autism spectrum disorder (ASD) include difficulties in grasping those aspects of communicated meaning that are not stated explicitly in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5, American Psychiatric Association 2013, pp. 85–86). A straightforward way for proponents of social motivation theory to account for pragmatic deficits in autism is to argue that lack of impetus to interact with others results in poor interactional experience, which, in turns, deprives autistics from a crucial opportunity to learn to interpret context-dependent, non-literal aspects of linguistic utterances. Reaching beyond what is literally said ultimately entails deriving assumptions about the correct speaker's meaning (which, incidentally, is one reason for introducing of the umbrella social (pragmatic) communication disorder in the DSM-5 (Lord et al. Reference Lord, Rutter, DiLavore, Risi and Gotham2012)). Insofar as all pragmatic processes inherently have a social component, one should expect lack of social motivation to uniformly impact all types of pragmatic skills.

However, recent research on pragmatics in autism reveals a rather different pattern. For instance, once receptive vocabulary is controlled for, autistics children and adults seem to understand metaphors (e.g., Hermann et al. Reference Hermann, Haser, Van Elst, Ebert, Müller-Feldmeth, Riedel and Konieczny2013; Norbury Reference Norbury2005). Autistic children and adults grasp requests that are cast in an indirect way, such as Your lunch is in your bag or Is it possible to open the door? (Deliens et al. Reference Deliens, Papastamou, Ruytenbeek, Geelhand and Kissine2018b; Kissine et al. Reference Kissine, De Brabanter and Leybaert2012; Reference Kissine, Cano-Chervel, Carlier, De Brabanter, Ducenne, Pairon, Delvenne and Leybaert2015). Individuals on the autism spectrum have also been consistently shown to derive non-logical, pragmatic interpretations of quantifiers, namely, interpreting some as some, but not all and X or Y as X or Y, but not both (Chevallier et al. Reference Chevallier, Wilson, Happé and Noveck2010; Pijnacker et al. Reference Pijnacker, Hagoort, Buitelaar, Teunisse and Geurts2009; van Tiel & Kissine Reference van Tiel and Kissine2018). Crucially, these latter interpretations are based on the exclusion of a more informative alternative (all for some and and for or), which, at the very least, entails awareness of the informativity level that can usually be expected in a verbal exchange.

Interestingly, some of the studies that reveal intact pragmatic abilities in ASD are co-authored by advocates of the social motivation theory (Chevallier et al. Reference Chevallier, Wilson, Happé and Noveck2010). Chevallier et al. (Reference Chevallier, Noveck, Happé and Wilson2011) even claim that pragmatic processes as complex as irony comprehension are preserved in ASD. These authors suggest that pragmatic – and, for that matter, mind-reading – skills may not be intrinsically deficient in ASD; however, they would not be spontaneously used, due to a diminished drive toward social interaction and poor orientation to social cues. An explicit prediction of this view is that, ceteris paribus, the performance on pragmatic tasks should depend on the strength of motivation to engage with the communicative partner (Chevallier et al. Reference Chevallier, Wilson, Happé and Noveck2010). That is, in experimental conditions where factors influencing motivation are kept constant, different pragmatic interpretations should be elicited at the same rate.

Importantly, in Chevallier et al. (Reference Chevallier, Noveck, Happé and Wilson2011)’s irony task, to choose between an ironic and a literal interpretations, participants could rely on the association between manifest incongruence with the literal meaning and a distinctive prosodic cue. Deliens et al. (Reference Deliens, Antoniou, Clin, Ostashchenko and Kissine2018a) designed an act-out task that, unlike Chevallier et al. (Reference Chevallier, Noveck, Happé and Wilson2011)’s forced-choice paradigm, makes it impossible to couple the ironic reading with a particular pattern of stimuli and requires genuinely reasoning about the speaker's intentions. In this paradigm, autistic participants experience strong difficulties in grasping irony (Deliens et al. Reference Deliens, Papastamou, Ruytenbeek, Geelhand and Kissine2018b, exp. 2). Crucially, however, in a task that requires deriving indirect request interpretations the same autistic participants who struggled with irony comprehension display a performance similar to that of neurotypicals (Deliens et al. Reference Deliens, Papastamou, Ruytenbeek, Geelhand and Kissine2018b, exp. 1). A similar picture emerges from van Tiel and Kissine (Reference van Tiel and Kissine2018). Their participants on the autism spectrum do not differ from neurotypicals in their pragmatic interpretation of simple quantified sentences, judging sentences of the form Some X's are Y as false in a context where all X's were Y – hence displaying a pragmatic, non-logical interpretation of some as some, but not all. However, the same participants do struggle with more complex “distributive” inferences; unlike neurotypicals, they tend to judge sentences of the form Each X is Y or Z as true even in contexts where all Xs were Y. Importantly, while pragmatic interpretations of some may be arrived at without genuinely making assumptions about the speaker's mental states, distributive inferences involve reasoning about what the speaker would have implied had she used the stronger alternative Each X is Y (or Each X is Z).

It is, therefore, likely that pragmatic processes that genuinely require adopting the speaker's point of view, such as irony or more complex pragmatic inferences, remain challenging for people on the spectrum. There is little reason to assume that autistic participants are less motivated in pragmatic tasks involving irony or distributive inferences than on those tapping indirect requests comprehension or the pragmatic reading of some. In fact, eye-tracking data or reaction times suggest no disengagement from the former tasks by participants on the autism spectrum (Deliens et al. Reference Deliens, Papastamou, Ruytenbeek, Geelhand and Kissine2018b; van Tiel & Kissine Reference van Tiel and Kissine2018). There are several possible explanations for the selective pragmatic impairment in ASD – for instance, in terms of impaired mind reading or executive dysfunction – that should be explored in future research. What is very unlikely is that poor social motivation would be the reason why autistics derive some context-dependent pragmatic meaning, but struggle with others.