Donders Institute for Brain, Cognition and Behaviour
Zoek in de site...

Theisis defense Irina Somanova (Donders Series 148)

12 May 2014

Promotor: Prof.dr. P. Hagoort

In search of conceptual representations in the brain: towards mind-reading

Throughout human history, people have always been intrigued by the idea of understanding thoughts from others without using speech or signs. In the 19th and 20th century mind reading was even studied by empirical psychologists. While this early work remained in the realm of pseudoscience, the idea has not become less attractive. Can humans acquire the ability to read minds? It remains highly unlikely that our sensory systems can be used to physically sense thoughts of others. Instead, however, a device could be constructed that monitors the neuronal activity in one’s brain, decodes it, and transforms it into a form that is meaningful to others. In the series of experiments, Irina Simanova, a PhD student in MPI Neurobiology of Language Department, explored the possibility to “read” mental states related to semantic processing from neuroimaging data. Simanova defends her thesis, "In Search of Conceptual Representations in the Brain: Towards Mind-Reading", on May 12 at 14.30, in the Radboud University Aula.

In the experimental studies Simanova focused on describing patterns of neuronal activity that reflect semantic processing. She attempted to identify neural patterns that are common across different modalities of perception (e.g. seeing, hearing, reading) and in speech production. Simanova collected neuroimaging data while participants perceived or thought about objects belonging to different semantic categories: animals or tools. She then applied machine learning classifiers to the data in order to extract activity patterns that are informative for the category.

The findings from the four experimental studies yield a consistent picture on the neural architecture for modality-independent semantic representations in the human brain, and have a number of important theoretical and methodological implications. First, the experiments show that the left posterior and temporal cortices are involved in conceptual processing, and neuronal circuits in these regions support the representation of semantic knowledge by integrating information originating from different input streams.

Second, the work by Simanova bridges the research on semantic memory in production, recall and perception. The experiments show that the brain areas that support semantic processing with different modalities of perception also underlie initial stages of word production: the retrieval in the absence of perceptual cues.

Finally, the results indicate that some aspects of semantic information (e.g. the semantic category) can be decoded from the measured brain signal. This finding encourages future research on decoding conceptual information during imagery and speech production.