SummaryCognitive flexibility is fundamental to adaptive intelligent behavior. Prefrontal cortex has long been associated with flexible cognitive function, but the neurophysiological principles that enable prefrontal cells to adapt their response properties according to context-dependent rules remain poorly understood. Here, we use time-resolved population-level neural pattern analyses to explore how context is encoded and maintained in primate prefrontal cortex and used in flexible decision making. We show that an instruction cue triggers a rapid series of state transitions before settling into a stable low-activity state. The postcue state is differentially tuned according to the current task-relevant rule. During decision making, the response to a choice stimulus is characterized by an initial stimulus-specific population response but evolves to different final decision-related states depending on the current rule. These results demonstrate how neural tuning profiles in prefrontal cortex adapt to accommodate changes in behavioral context. Highly flexible tuning could be mediated via short-term synaptic plasticity.

There has been considerable debate as to whether the hippocampus and perirhinal cortex may subserve both memory and perception. We administered a series of oddity tasks, in which subjects selected the odd stimulus from a visual array, to amnesic patients with either selective hippocampal damage (HC group) or more extensive medial temporal damage, including the perirhinal cortex (MTL group). All patients performed normally when the stimuli could be discriminated using simple visual features, even if faces or complex virtual reality scenes were presented. Both patient groups were, however, severely impaired at scene discrimination when a significant demand was placed on processing spatial information across viewpoint independent representations, while only the MTL group showed a significant deficit in oddity judgments of faces and objects when object viewpoint independent perception was emphasized. These observations provide compelling evidence that the human hippocampus and perirhinal cortex are critical to processes beyond long-term declarative memory and may subserve spatial and object perception, respectively.

A series of five experiments investigated the relationship between object memory and scene memory in normal and fornix-transected monkeys. An algorithm created formally defined background and objects on a large visual display; the disposition of some particular objects in particular places in a particular background constitutes a formally defined scene. The animals learned four types of discrimination problem: (1) object-in-place discrimination learning, in which the correct (rewarded) response was to a particular object that always occupied the same place in a particular unique background, (2) place discrimination learning, in which the correct response was to a particular place in a unique background, with no distinctive object at that place, (3) object discrimination learning in unique backgrounds, in which the correct response was to a particular object that could occupy one or the other of two possible places in a unique background, and (4) object discrimination learning in varying backgrounds, in which the correct response was to a particular object that could appear at any place in any background. The severest impairment produced by fornix transection was in object-in-place learning. Fornix transection did not impair object discrimination learning in varying backgrounds. The results from the other two types of learning task showed intermediate severity of impairment in the fornix-transected animals. The idea that fornix transection in the monkey impairs spatial memory but leaves object memory intact is thus shown to be an oversimplification. The impairments of object memory in the present experiments are analogous to the impairments of episodic memory seen in human amnesic patients.

Aspiration lesions of the amygdala were found previously to produce a severe impairment in visual discrimination learning for auditory secondary reinforcement in rhesus monkeys (Gaffan and Harrison, 1987). To determine whether excitotoxic amygdala lesions would also produce this effect, we trained four naive rhesus monkeys on the same task. The monkeys were required to learn 40 new visual discrimination problems per session in a situation in which visual choices were guided by an auditory secondary reinforcer that had been previously associated with food reward. Bilateral excitotoxic lesions of the amygdala had no effect on the rate of learning visual discrimination problems for auditory secondary reinforcement. We also tested the amygdalectomized monkeys on a reinforcer devaluation task and compared their performance with a group of three normal monkeys. The monkeys first learned to discrimi-nate 60 pairs of objects, baited with two different food rewards. Each of the food rewards was then devalued by selective satiation in two separate experimental sessions. Normal controls tended to avoid displacing objects that covered the devalued food to a significantly greater degree than did the amygdalectomized monkeys, indicating that the excitotoxic amygdala damage interfered with reinforcer devaluation effects. Our results are consistent with the idea that the amygdala is necessary for learning the association between stimuli and the value of particular food rewards; however, the amygdala is not necessary for maintaining the value of secondary reinforcers, once they have been learned.

There has been considerable debate as to whether structures in the medial temporal lobe (MTL) support both memory and perception, in particular whether the perirhinal cortex may be involved in the perceptual discrimination of complex objects with a large number of overlapping features. Similar experiments testing the discrimination of blended images have obtained contradictory findings, and it remains possible that reported deficits in object perception are due to subtle learning in controls, but not patients. To address this issue, a series of trial-unique object "oddity" tasks, in which subjects selected the odd stimulus from a visual array, were administered to amnesic patients with either selective bilateral damage to the hippocampus or more extensive damage to MTL regions, including the perirhinal cortex. Whereas patients with damage limited to the hippocampus performed similarly to controls on all conditions, patients with perirhinal damage were significantly impaired when the task required discrimination between objects with a large number of features in common. By contrast, when the same stimuli could be discriminated using simple visual features, patients with perirhinal damage performed normally. These results are consistent with a theoretical view which holds that rostral inferotemporal cortical regions, including perirhinal cortex, represent the complex conjunctions of stimulus features necessary for both perception and memory of objects.