The nature of the pragmatic impairment in autism spectrum disorders


Although it has long been known that deficits in language and communication are defining characteristics of the autism spectrum disorder (ASD) phenotype, the precise nature of these difficulties, their developmental trajectory, and their underlying neural mechanisms are not known. Understanding language entails much more than simply decoding the literal meaning of each sentence: our interpretation of each utterance is powerfully shaped by our knowledge of the intent of the speaker, the linguistic and social context of the utterance, and our general world knowledge. This ability to exploit speaker intent and background knowledge to go beyond the literal meaning of the sentence is at the core of the communication deficit in autism. We will use a synergistic combination of functional neuroimaging (fMRI), behavioral, developmental and computational approaches to richly characterize the nature of the pragmatic impairment in autism spectrum disorders. Ev Fedorenko and Nancy Kanwisher are the managing PIs for the project as a whole.

Component Projects:

Characterizing the contributions of three neural systems — the core language system, the Theory of Mind system, and the cognitive control system — to pragmatic reasoning (Ev Fedorenko and Rebecca Saxe)

Pragmatic understanding requires three component mental abilities to work in concert: a) basic language processing, b) aspects of social cognition (especially Theory of Mind), and c) domain-general executive abilities (like working memory and cognitive control). Core language abilities are necessary for deriving the literal interpretation of an utterance. Theory of Mind is critical for taking into account the perspective of the speaker, which is essential for successful communication. And working memory is necessary for maintaining multiple, often conflicting representations over time, as the conversation unfolds. We here propose to investigate the division of labor among these three cognitive systems – each supported by a distinct large-scale neural network – during pragmatic reasoning, and to characterize the nature of the pragmatic impairment in autism spectrum disorders. Our proposal has two specific aims.

Aim 1: is to characterize the contributions of the language system, the Theory of Mind (ToM) system, and the cognitive control system to pragmatic reasoning.
Aim 2: is to relate key neural measures to behavioral performance on tasks tapping pragmatic competence, including tasks used in Project 2.

Investigating the nature of the pragmatic impairment in individuals with ASD using a set of robust communicative paradigms (Edward Gibson)

Although individuals with autism spectrum disorders (ASD) appear to be able to extract the core literal meaning from linguistic utterances, they are impaired at interpreting those meanings in the social and linguistic contexts in which they occur and to take into account the mental states of the speakers. This project aims to shed light on the nature of this pragmatic impairment using a set of novel robust paradigms developed in the last several years in the field of psycholinguistic research. In particular, we propose a series of behavioral experiments to investigate pragmatic influences in three areas of language use, to be carried out in ASD individuals and healthy control participants. These experiments will investigate three of the most important components of language structure: words, sound structure across utterances (prosody), and sentence structure (syntax).

Aim 1: Lexical production and comprehension: To investigate syntactic and semantic surprisal in sentence completions.
Aim 2: Prosody production and comprehension: To investigate the production and comprehension of prosodic cues in different contexts.
Aim 3: Syntactic production: To investigate word order in a task where participants gesture meanings of simple events.

A computational and behavioral investigation of our reasoning about others’ utility functions in developing and mature individuals with ASD and neurotypical participants (Laura Schulz and Josh Tenenbaum)

The goal of this project is to deepen our understanding of how people estimate other people’s utility function (their costs and rewards), and to explore the kinds of social and communicative inferences (e.g., about others’ goals, desires, beliefs, and moral status) that knowledge of these utility functions supports. To do this, we will develop and implement a computational model of how neurotypical adults make these inferences. We will then investigate the ways that differences in the cost function of agents influence the inferences we make about other people. To understand both typical and clinical development, we will test the predictions of the model on four different populations: typically developing adults, typically developing preschoolers, adults with autism, and preschoolers with autism.

Aim 1:
To develop a computational model of how we infer both agent-invariant and agent-specific costs of action.
Aim 2:
To test, in typical adults and children as well as adults and children with autism, the ways that information about agents’ utility function guides the representation of efficient action and reasoning about social evaluations.