From British Associationism to the Hippocampal Cognitive Map: A Personal View From a Ringside Seat at the...

The mandate for this special issue of Hippocampus was to provide a few examples of one's own work in a relatively personal context. Accordingly, I will discuss some of my own work here, but will also provide a broader arc of ideas and discoveries within which the efforts of myself and many others have taken place. This history begins with the associationists, who proposed that the human mind could be understood, in part, as a compounding of simple associations between contiguously occurring items and events. This idea was taken up by the behaviorist traditions, which made significant progress toward refining this simple idea. Subsequently, as interest turned toward neural mechanisms, Donald Hebb provided a foundational proposal for how synaptic changes could provide for this associative learning. The associationist view was, however, challenged by gestaltists who took a more wholistic, cognitive approach. Stunning support was provided for Tolman's cognitive map idea with the discovery of place cells in the hippocampus, and the subsequent treatise provided in O'Keefe and Nadel's The Hippocampus as a Cognitive Map. I propose that, ultimately, this associationist versus cognitive debate was settled by the development of the Parallel Distributed Processing (PDP) approach, which incorporated Hebbian synapses into large, neural-like networks, which could accomplish complex cognitive tasks. My own work took place within the framework provided by O'Keefe and Nadel. One aspect of my work followed Jim Ranck's discovery of Head Direction cells. Tad Blair and others in my lab traced a brainstem circuit, which we proposed could explain the origins of the directional code. In other work, I investigated cells in the subicular region. These provided a contrast to the hippocampal place cells in that each subicular cell kept the same spatial pattern across different environments, whereas the hippocampal cells formed a different map for each context.