Substrates of Recognition Memory in Mice

Recognition memory is a fundamental cognitive process that declines with age. Animal models of recognition memory are thus central to research into the neural basis of recognition and its age-related decline. Such work typically uses the novel object recognition (NOR) task, in which a previously exposed item elicits less exploration than a novel one. Although NOR is widely used, theoretical interpretation of NOR performance is controversial. Translational theories of recognition are typically based on the human distinction between recollection and familiarity. This approach is not only relatively loose in its predictions, but is also being increasingly questioned on theoretical and methodological grounds. Thus we have argued (e.g. Robinson & Bonardi, 2015) for a different approach. We adopt an associative account of recognition memory based on animal work (SOP: Sometimes Opponent Process; Wagner, 1981) to explain performance on the NOR task.

According to this theory, every stimulus comprises a set of constituent elements that in its absence are inactive. When the stimulus occurs some of its elements are activated into a state of primary activation called a1, corresponding to the stimulus being in the centre of attention. a1 has only limited capacity, and elements in a1 decay rapidly into a secondary activation state a2, where they command less attention; elements in a2 slowly become inactive again (Fig.1 black arrows).

Imagine hearing a doorbell: initially the unexpected ring occupies your attention (a1), but this quickly gives way to a less prominent awareness that the bell has rung (a2), until eventually you stop thinking about it altogether (inactive). Critically, this is a one-way cycle: an element in a2 must become inactive again before it can re-enter a1; moreover, an element in a1 elicits a stronger response than one in a2. In other words every stimulus presentation is followed by a refractory period during which that same stimulus elicits a weaker response (because many of its elements are still in a2 and so cannot immediately re-enter a1). Thus you jump when the bell first rings, but less so if it rings again immediately. This transient ability of a stimulus to reduce its own impact the next time it occurs is termed self-priming (Sp).

A second type of priming is retrieval-priming (Rp), and this relies on associative learning. If one stimulus is quickly followed by another, they become associated (as both have elements in a1). Once associated, presenting the first stimulus associatively retrieves the second, sending its elements directly into a2 (Fig.1 grey arrow). When the retrieved stimulus actually occurs responding to it is reduced (as many of its elements are already in a2; thus, as in Sp, fewer of them can enter a1). In short, if you expect the doorbell to ring it will startle you less. This reduction in the impact of a retrieved stimulus is retrieval-priming (Rp).

This theory asserts that recognition is a product of both Sp and Rp, and that these two processes - not familiarity and recollection - are the building blocks of recognition.
The aims of this proposal are to test the two core assertions of this account of recognition memory, and to refine existing tasks to provide pure measures of Sp and Rp, by devising variants that rule out alternative explanations of performance. These aims are achieved by the following objectives:
A) test the fundamental prediction that associations underlie recognition effects;
B) test the prediction that novel object recognition stems from two independent mechanisms: self-priming and retrieval-priming from associations between the object and the arena cues.
The aim of this PhD is to conduct a systematic test of this associative account of recognition memory in a series of experiments with mice.