Neural Correlates Underlying Binding of Information in Autism: Preliminary Results

Background:  Working memory (WM) is a limited-capacity system devoted to temporary storage and short-term handling of information. Within the different components of working memory, the ability to combine various information, referred to as binding, is crucial for episodic memory and learning processes. There are two types of binding: the automatic form, for which associations between pieces of information are made unconsciously (information being presented already bound together), and the controlled form, for which these associations require a conscious effort. These two forms of binding are based on separate neural substrates: automatic binding is based on the functioning of the hippocampus, while controlled binding is based on the communication between the hippocampus and prefrontal cortex. Some studies have suggested preserved automatic binding capabilities in autistic individuals, while other studies have shown controlled binding difficulties. This profile of performance could result from alterations of the neural circuit connecting the prefrontal cortex and hippocampus.

Objectives:  To investigate the cognitive and brain mechanisms underlying both automatic and controlled forms of information binding in autism, using functional magnetic resonance imaging (fMRI).

Methods:  13 autistic participants (diagnosed by a multidisciplinary team with ADI-R and ADOS-G) and 15 typically developed participants were recruited from the research database of a specialized clinic for autism diagnosis. Participants were matched on gender, IQ and age (18-40 years old). In an MRI scanner, participants performed the working memory experimental task, which consisted of the presentation of words and ellipses (indicating a spatial position) of different colors, to memorize. In the automatic condition, words were presented directly into ellipses. In the controlled condition, the words were presented in the center of the screen, separated from ellipses. Participants had to associate same color words and ellipses. Stimuli were presented for 3 seconds and participants had to maintain information for 10 seconds in WM. After this delay, a word was presented in a spatial position and participants had to decide whether the word/spatial position association was correct or not.

Results: Participants had significantly lower accuracy in controlled binding than in automatic binding, with no between-group differences. During encoding phase, typical participants presented greater levels of activity in the middle frontal gyrus for controlled binding, and in dorsolateral prefrontal cortex for automatic binding, compared to autistic participants. During maintenance phase, autistic participants showed greater levels of activity in the left median frontal gyrus, hippocampus and parahippocampal gyrus, for controlled binding, and in the parahippocampal gyrus and left inferior gyrus for automatic binding, compared to typical participants. 

Conclusions:  Despite similar performances, autistic and typical participants presented different patterns of cerebral activity. When encoding information, the different level of activity in prefrontal cortex suggests the use of different encoding strategies in the two groups. Typical participants probably used a more active strategy of manipulation of information to remember. However, when maintaining information in WM, greater activity in the hippocampus and prefrontal regions was found in autistic participants, suggesting more effortful strategies employed by autistic participants.