Funded Grants

Seeing is remembering: Probing the common representations underlying memory and perception

Memory provides an anchor to the past that enables one to make sense of the present and project oneself into the future. When the ability to form new memories is lost, as occurs in Alzheimer’s disease or following brain damage, the consequences are devastating. For more than fifty years, scientists have known that damage to the medial temporal lobes (MTL) – a set of heavily interconnected structures deep inside the brain – causes amnesia. However, despite decades of intense research, the means by which different structures in the MTL contribute to memory acquisition and loss remain the subject of intense debate.

The most influential view of memory organization holds that structures in the MTL form a dedicated long-term memory system, with no role in other cognitive processes such as perception, attention, or working memory. This view is consistent with a popular paradigm in cognitive neuroscience, in which the brain is understood in terms of neatly segregated modules that map onto broad psychological constructs. Within this framework, it is argued that separate brain systems account for distinct cognitive processes, like “declarative memory,” “priming,” or “recollection.” Here, I suggest that this view is neither the most parsimonious nor the most compelling account of the data. Though MTL structures are essential for memory, findings from my laboratory indicate that these MTL structures do, in addition, play a critical role in perception – a finding that is fundamentally incompatible with the notion of a dedicated MTL memory system specifically, or cognitive modules more generally.

In contrast to a modular view of memory, I argue that the functions of brain regions can be better understood in terms of the representations they contain and the computations they perform. I suggest that the MTL should be considered in terms of a continuous, hierarchical account of cognitive processing that begins in the ventral visual stream, a processing pathway known to be critical for the perception of objects. I propose that the perirhinal cortex (PRC) – a key structure of the putative MTL memory system – sits at the apex of this ventral visual perceptual stream, representing the complex conjunctions of features that comprise objects. These complex “object-level” PRC representations allow for fine-grained distinctions across objects, which becomes especially critical under conditions of high visual interference when differentiation of similar items among competing alternatives is required. Guided by computational modelling complemented with neuropsychology and neuroimaging, we will provide a unified and mechanistic account of why MTL damage affects both mnemonic and perceptual processes in humans.

This project stands to redefine current conceptions of memory disorders by shifting the focus from explaining deficits in terms of impaired cognitive modules (e.g., for long-term memory or perception) to explaining deficits in terms of impoverished neural representations. This idea has profound implications for cognitive rehabilitation and lays the foundation for new therapeutic techniques in memory disorders.