A Roadmap for Understanding Memory: Decomposing Cognitive Processes into Operations and Representations

Abstract

Thanks to patients Phineas Gage and Henry Molaison, we have long known that behavioral control depends on the frontal lobes, whereas declarative memory depends on the medial temporal lobes. For decades, cognitive functions – behavioral control, declarative memory – have served as labels for characterizing the division of labor in cortex. This approach has made enormous contributions to understanding how the brain enables the mind, providing a systems-level explanation of brain function that constrains lower-level investigations of neural mechanism. Today, the approach has evolved such that functional labels are often applied to brain networks rather than focal brain regions. Furthermore, the labels have diversified to include both broadly-defined cognitive functions (declarative memory, visual perception) and more circumscribed mental processes (recollection, familiarity, priming). We ask whether a process – a high-level mental phenomenon corresponding to an introspectively-identifiable cognitive event – is the most productive label for dissecting memory. For example, the process of recollection conflates a neurocomputational operation (pattern completion-based retrieval) with a class of representational content (associative, high-dimensional, episodic-like memories). Because a full theory of memory must identify operations and representations separately, and specify how they interact, we argue that processes like recollection constitute inadequate labels for characterizing neural mechanisms. Instead, we advocate considering the component operations and representations of mnemonic processes in isolation, when examining their neural underpinnings. For the neuroanatomical organization of memory, the evidence suggests that pattern completion is recapitulated widely across cortex, but the division of labor between cortical sites can be explained by representational content.

The locus of recognition memory signals in human cortex depends on the complexity of the memory representations

“…In this study, we tested two hypotheses: first, that the cortical locus of recognition memory signals depends on the complexity of the to-be-remembered content; and second, that the location of memory signals tracks the location of the optimal neural code for the representations being retrieved. These hypotheses entail that recognition memory does not always depend on the MTL (contrasting with dominant memory theories, e.g., Brown & Aggleton, 2001;Davachi, 2006;Diana et al, 2007;Eichenbaum et al, 2007;Norman & O'Reilly, 2003;Squire & Zola-Morgan, 1991;Yonelinas, Ranganath, Ekstrom, & Wiltgen, 2019), but can rely on extra-MTL sites when the representations being retrieved are sufficiently simple or low-dimensional (Cowell et al, 2019;Cowell, Bussey, & Saksida, 2010a;Graham et al, 2010).…”

“…However, an important legacy of the Swiss army knife view is that most research questions still center on a cognitive function − e.g., episodic memory retrieval, decision-making, or selective attention − and ask which brain region, distributed network, or neural mechanism supports it. That is, cognition is still carved into subdomains that correspond to functional "tools" (Cowell, Barense, & Sadil, 2019).…”

“…But many such models still explain brain regions' contributions via a combination of function and content, e.g., claiming the hippocampus underpins recollection (a mnemonic function) of contextual details (a type of content) (Bastin et al, 2019;Davachi, Mitchell, & Wagner, 2003;Eichenbaum, Yonelinas, & Ranganath, 2007;Staresina, Cooper, & Henson, 2013). We, and others, have suggested that "content" is more than just a label to be used in combination with mnemonic functions or processes, such as recollection; instead, it may be the primary principle driving how memory and perception map onto the brain at the systems level (Bussey & Saksida, 2002;Cowell et al, 2019;Graham, Barense, & Lee, 2010;Nadel & Peterson, 2013).…”

How displacement might have evolved

“…More specifically, this region has been shown to be closely related to MTT, as well as the encoding of spatial information (Eichenbaum, 2015;Schiller et al, 2015). The hippocampus serves not only as a storage device that receives multisensory input, but also as a computational device combining information of assorted nature, including temporal and contextual information (Cowell et al, 2019;Shimamura, 2010).…”

Content-specific vulnerability of recent episodic memories in Alzheimer's disease

“…New accounts consider interactions between these process- and content-oriented approaches to understand episodic memory function and acknowledge that systems storing specific types of representations may be shaped by computational operations that are executed in certain subregions (e.g. Bastin et al, 2019 ; Cowell et al, 2019 ; see also Ekstrom and Yonelinas, 2020 ). Future research may show how certain computations require and shape specific representations (e.g.…”