Rebecca Saxe - US grants

Our ability to reason about the thoughts of other people is the bedrock of our daily lives. Adults must think about others' thoughts constantly, from simple conversations to bluffing in a poker game. Even one-year-old babies learning their first words rely upon sophisticated inferences about the intentions of the speaker. Thinking about thoughts is ubiquitous, but it presents a serious cognitive and computational challenge. Prior research by Dr. Rebecca Saxe of MIT suggests that typical human adults achieve this feat partly by using a dedicated neural mechanism -- a group of brain regions that are recruited for thinking about thoughts. This CAREER award will pursue the implications of these results: How do specialized brain regions for thinking about thoughts develop in childhood? What role do these regions play in the social cognitive deficits observed in Autism? How is the development of these brain regions affected by changes in childhood experiences, such as delayed access to language? The proposed studies will investigate these questions by using behavioral tasks to measure development in children's ability to think about others' thoughts, and functional neuroimaging to measure concomitant development in the brain. The experiments will compare task performance and neural responses in typically developing children, children diagnosed with Autism, and children who experienced delayed exposure to language (e.g., deaf children of non-signing parents).

The research proposed here will generate new insights both within developmental cognitive neuroscience, concerning the structure and function of a fundamental domain of human cognition, and across disciplinary boundaries, addressing key questions in developmental psychology and neuroscience. The project may also help guide the translation of this basic research into interventions for children who are socially at risk. Additionally, the project will support the development of an elementary school neuroscience curriculum to bring the generated knowledge about children's developing brains back to the children themselves.

DESCRIPTION (provided by applicant): Humans think constantly about one another's thoughts, for example, in order to communicate, to teach, to learn from, to cooperate with, to compete with, and to deceive one another. This ability to fluently infer what others are thinking i impaired in individuals with Autism Spectrum Disorders (ASD), but the neural basis of this impairment is poorly understood. In particular, brain regions involved in thinking about other people's thoughts are not reliably smaller or less active in individuals with ASD. This proposal will test the possibilities that (i) reduced magnitude of activity, (ii) disorganization of the patern of activity, and/or (iii) altered connectivity of these brain regions that leads to social impairmets in ASD. We will test these hypotheses about the function of ToM regions in adults with ASD, in three key areas of ToM: understanding others' thoughts that are relevant for moral judgments (e.g. intentional vs. accidental harm), for inferring emotions, and for effective communication. Since ASD is a developmental disorder, it is critical to understand the developmental trajectory of these brain regions as well as their end state. Consequently, we will also test these hypotheses in children aged 5-12 years. We will also test whether the selectivity or pattern of activation in ToM brain regions changes over development. PUBLIC HEALTH RELEVANCE: Humans think constantly about one another's thoughts, for example, in order to communicate, to teach, to learn from, to cooperate with, to compete with, and to deceive one another. The ability to fluently infer what others are thinking is impaired in individuals with Autism Spectrum Disorders (ASD), but the neural basis of this impairment is poorly understood. This proposal will test the hypotheses that brain regions typically involved in thinking about other people's thoughts (i) are smaller or less active, (ii) have a disorganized pattern of activity, or (iii) have inappropriate connectivity, which leads to social impairments in adults and children with ASD.

Infants' limited ability to communicate makes it difficult to know what they are thinking. Researchers often measure babies' visual attention as a way to infer their understanding and interest. However, knowing that infants prefer to look at one display versus another does not necessarily indicate why they prefer it. Some attentional preferences are driven by how much the infant seems to like the stimulus, referred to as its reward value. Other preferences are driven by infants' desire to take in new information and learn more about the world. The goal of the proposed research is to study different patterns in babies' brain activity to determine which of these two reasons, reward value or information value, explains their attentional preferences.

The proposed experiments will use functional near-infrared spectroscopy (fNIRS) to measure hemodynamic responses to neural activity in medial and dorsolateral prefrontal cortex while infants watch and listen to two different speakers. In some experiments, the speakers will differ in how friendly and infant-directed their speech is (i.e., social reward), while in others they will differ in how richly structured the content of their speech is (i.e., information value). All experiments will end by testing to which speaker the infants prefer to attend and/or how much they have learned about the presented speech patterns. These experiments will test the hypotheses that activation in medial prefrontal cortex (MPFC) is associated with social reward value and activation in dorsolateral prefrontal cortex (DLPFC) is associated with information value. Moreover, they will assess the extent to which activation in MPFC and DLPFC during stimulus presentation predict subsequent social preference and statistical learning, respectively. The project will also involve scientific training of students from underrepresented groups and underprivileged backgrounds, and conducting information sessions for new parents about infant brain development.

Project Summary In the last two decades, functional neuroimaging has providing a stunning window into the function of the human brain. When a person looks around the world, for example, systematic patterns of activity emerge in the cortex, related to both visual properties (e.g. the size or position) and also conceptual properties (e.g. object versus human versus natural scene) of the experience. These patterns of cortical activity are remarkably consistent across different adults, but very little is known about when, or how, they develop. Understanding the timing and mechanism of typical brain development is a necessary foundation for identifying the cause of neuro-developmental disorders that affect cognitive development, and for assessing the ef?cts of early rehabilitation. One reason these questions remain unanswered is the limited methods available for neuroimaging in awake human infants. To measure functional responses in the whole cortex with high spatial resolution, the ideal technique is functional magnetic resonance imaging (fMRI); however, it is extremely challenging to acquire high-quality fMRI data from infants while they are awake. One challenge is participant motion. Moving a millimeter is enough to destroy a MR image. Infants cannot be instructed to lie still, and cannot be held tightly because their skulls are soft (and because they must be comfortable during the experiment). The ?rst aim of the proposed research is to address the challenges of fMRI in infants, and collect high quality data from awake human infants while they watch dynamic, bright colored, infant-friendly movies. Technical innovations necessary to achieve this aim include novel ?exible experimental designs, custom-built coil sized for an infant's head, a special protocol for scanning the infant with a parent in the MRI, custom-written quiet MRI sequences, and special procedures for data analysis. The second aim is to compute and compare the responses to 60 blocks of movies (including children's faces and bodies, toys, natural scenes, and textures) in twenty full-term infants aged 4-8 months and in adults. The patterns of responses across movies can also be used to compare the predictions of alternative theories of both initial state and learning mechanisms in infant cortex. Finally, when cortex changes over development, a fundamental question is whether this change is driven by learning from visual experience, or by maturation of the biological mechanism. Because visual experience is extremely limited in utero, a natural dissociation of these factors occurs in the case of moderately preterm birth. Measuring the development of cortical responses in full-term and moderately preterm infants will also provide an important foundation for studies of infants born very preterm, who are at high risk for neurological and developmental disorders.