Behavioral and Neural Mechanisms of Latent Extinction: A Historical Review

The present paper provides a comprehensive review of latent extinction. In maze learning situations, latent extinction involves confining an animal to a previously reinforced goal location without food. When returned to the starting position after latent extinction, the animal typically shows a response decrement. Such findings have suggested that latent extinction is sufficient to invoke extinction learning, despite the animal having been prevented from making the original response. The majority of research on latent extinction was conducted between 19491980 and focused on what is being learned during the latent placements. Stimulus-response (S-R) theorists attempted to explain latent extinction via novel S-R mechanisms, namely, the fractional anticipatory response (rG). However, research did not support the role of rG in latent extinction. Cognitive expectancy theorists provided a simpler, more adequate explanation for latent extinction, more consistent with the data. Specifically, latent extinction might instill a change in expectation (i.e., animals learn to expect absence of reinforcement). Evidence also suggests that latent extinction involves place learning mechanisms and is sensitive to modulation via certain experimental factors. More recent work has uncovered some of the neural mechanisms of latent extinction. The hippocampus is critically involved in latent extinction, whereas other brain regions typically implicated in regular "response extinction" in the maze, such as the dorsolateral striatum, are not required for latent extinction. Similar to other kinds of learning, latent extinction requires NMDA receptor activity, suggesting the involvement of synaptic plasticity. Consistent with a multiple memory systems perspective, research on latent extinction supports the hypothesis that extinction learning is not a unitary process but rather there are different kinds of extinction learning mediated by distinct neural systems.