Key words: Episodic Memory; Hippocampus; Binding; Recruitment Abstract
The memorization of events and situations (episodic memory) requires the rapid formation of a memory trace consisting of several functional components. A computational model is described that demonstrates how a transient pattern of activity representing an episode can lead to the rapid recruitment of appropriate circuits as a result of long-term potentiation within structures whose architecture and circuitry match those of the hippocampal formation, a neural structure known to play a critical role in the formation of such memories. 1 Introduction
We remember our experiences in terms of events and situations that record who did what to whom where and when, or describe states of affairs wherein multiple entities occur in particular configurations. This form of memory is referred to as episodic memory  and it is known that the hippocampal system (HS) serves a critical role in the formation of such memories [10, 21, 4, 22].
A number of researchers have proposed models to explain how the HS subserves the episodic memory function. These include macroscopic system-level models that attempt to describe the functional role of the HS as well as more detailed computational models that attempt to explicate how the HS might realize its putative function (e.g., [11, 14, 22, 8, 13, 12]. While our understanding of the HS and its potential role in memory formation and retrieval has been enhanced by this extensive body of work, several key representational problems associated with the encoding of episodic memories have remained unresolved. In particular, most existing computational models view an item in episodic memory as a feature vector or as a conjunction of features. But for reasons summarized below, such a view of episodic memory is inadequate for encoding events and situations (also see [16, 18]). First, events are relational objects, and hence, cannot be encoded as a conjunction of features. Consider an event E1 described by “John gave Mary a book in the library on Tuesday”. This event cannot be encoded by simply associating “John”, “Mary”, “a Book”, “Library”, “Tuesday” and “give” since such an encoding would be indistinguishable from that of the event “Mary gave John a book in the Library on Tuesday”. In order to make the necessary distinctions, the memory trace of an event should specify the bindings between the entities participating in the event and the roles they play in the event. For example, the encoding of E1 should specify the following role-entity bindings: (hgiver=Johni, hrecipient=Maryi, hgive-object=a-Booki, htemporal-location=Tuesdayi, hlocation=Libraryi). Second, the memory trace of an event should be responsive to partial cues, but at the same time, it should be capable of distinguishing between the memorized event and other highly similar events. For example, the memory trace of E1 should respond positively to a partial cue “John gave Mary a book,” but not to a cue “John gave Susan a book in the library on Tuesday” even though the latter event shares all, but one, bindings with E1. Third, during retrieval, the memory trace of an event should be capable of reactivating the bindings associated with the event so as to recreate an active representation of the event. In particular, the memory trace of an event should support the retrieval of specific components of the event. For example, the memory trace of E1 should be capable of differentially activating John in response to the cue “Who gave Mary a book?” It can be shown that in order to satisfy all of the above representational requirements, the memory trace of an event must incorporate functional units that serve as binding-detectors, binding-error-detectors, binding-error-integrators, Figure 1: Summary of major pathways interconnecting the components of the hippocampal system. Note the multiple pathways from EC to CA1, CA2, CA3, and SC. Also note the backprojections from CA1 and SC to EC....
References:  D.G. Amaral and M.P. Witter, Hippocampal Formation, in: G. Paxinos, ed., The Rat Nervous System, second,
edition, (Academic Press, London, 1995) 443-493.
Nature 361 (1993) 31-39.
 J.J. Chrobak and G. Buzsaki, Gamma Oscillations in the Entorhinal Cortex of the Freely Behaving Rat. J. Neurosci.,
18 (1998) 388-398.
 N.J. Cohen and H. Eichenbaum,Memory, Amnesia, and the Hippocampal System (M.I.T. Press, Cambridge,Massachusetts,
 H.M. Duvernoy, The Human Hippocampus. (J.F. Bergmann,Munich, 1988).
 J.A. Feldman, Dynamic connections in neural networks. Bio-Cybernet. 46 (1982) 27-39.
 M.E. Hasselmo, B.P. Wyble and G.V. Wallenstein, Encoding and retrieval of episodic memories: Role of cholinergic
and GABAergic modulation in the hippocampus, Hippocampus 6 (1996) 693-708.
 R.T. Knight, Contribution of human hippocampal region to novelty detection. Nature 382 (1996) 256-259.
 J. O’Keefe and L. Nadel, The hippocampus as a cognitive map (Clarendon Press, Oxford, 1978).
 D. Marr, Simple memory: a theory for archicortex, Phil. Trans. R. Soc. B 262 (1971) 23-81.
 J.M.J. Murre, TraceLink: a model of amnesia and consolidation of memory. Hippocampus 6 (1996) 674–684.
 R.C. O’Reilly and J.L. McClelland, Hippocampal Conjunctive Encoding Storage, and Recall: Avoiding a Tradeoff,
Technical Report PDP.CNS.94.4, Carnegie Mellon University, Pittsburgh, PA, 1994.
Psych Rev. 99 (1992) 268-305.
 L. Shastri, Semantic Networks: An evidential formalization and its connectionist realization, (MorganKaufamnn,
Los Altos, CA, 1988).
 L. Shastri, Recruitment of binding and binding-error detector circuits via long-term potentiation, Neurocomputing
26-27 (1999) 865-874.
18 (1995) 555-586.
Rev. 99 (1992) 195-231.
 A. Treves and E.T. Rolls, Computational analysis of the role of the hippocampus in memory. Hippocampus 4
 E. Tulving, Elements of Episodic Memory (Clarendon Press, Oxford, 1978).
 C. von der Malsburg. Am I thinking assemblies? in: G. Palm and A. Aertsen, eds., Brain Theory (Springer-
Verlag, Berlin, 1986).
Ottersen, eds., Progress in Brain Research: Understanding the brain through the hippocampus (Elsevier Science,
Amsterdam, 1990) 13-36.
Please join StudyMode to read the full document