Steven P. Wise - US grants

The Section on Neurophysiology studies the frontal cortex and related parts of the brain. The present project was premised on the idea that making decisions about what to do and when to do it requires three distinct neural processes: comparison of the relative value of alternative actions (known as valuation or evaluation), selection of the alternative with the highest current value, and inhibition of rejected alternatives. Because the ability to evaluate, select, and inhibit a vast array of potential thoughts and actions is so fundamental to everyday life, dysfunction in any of these decision-making processes has potentially wide-ranging and deeply debilitating effects. Thus, it is not surprising that dysfunction of these three neural processes has been implicated in the thought disorders characteristic of schizophrenia, the deficits at the root of attention deficit hyperactivity disorder (ADHD), and the inability to willfully control thoughts and behavior in obsessive-compulsive disorder (OCD). Human brain imaging and clinical neuropsychological studies have provided evidence for a frontal-lobe contribution to all three decision-making processes, with much emphasis on the role of the prefrontal cortex. Currently, however, we lack a systems-level understanding of how these areas and specific receptor pathways work together to achieve these three fundamental processes of decision making: valuation, behavioral inhibition and behavioral selection. Accordingly, we hypothesized that the orbital prefrontal cortex (PFo) and both the core and shell of the nucleus accumbens, two parts of the basal ganglia, suppress a prepotent tendency to choose the largest of two values, but that they do so in different ways, some related to valuation, others to general or specific inhibition, and others to affirmative selection of a goal or response. If this is the case, they we predict that neurons in these areas should differ when subjects decide what to do based on a prepotent response rule (select the largest value), as opposed to the antithesis of that rule (select the smallest of two values). The foundation for this project is the reversed contingency task. We previously showed that rhesus monkeys can learn this task, in contradiction of previously published results (Murray et al., 2005). In the present neurophysiological part of the project, a red cue signaled that the payoff contingency will be the receipt of four identical food items for choosing four items (the prepotent task) and a blue cue signaled the contingency will be the receipt of only one food item for choosing four, along with the receipt of four food items for choosing one (the reversed contingency task). We used a moveable, multi-electrode drive, fitted to a newly designed drive base to record up to 16 areas simultaneously, typically targeting 5-10 areas per recording session. Extracellular recordings of single neurons, eye movements, and EMGs were analyzed. In a preliminary analysis, we found that 235 of 458 neurons, 51% of the sample throughout the frontal cortex and striatum, showed a statistically significant difference between the reversed contingency and prepotent tasks. In control tasks, 198 of 432 (45%) neurons showed a significant relationship with the amount of a self-imposed delay, and 167 of 362 cells (46%) reflected the amount of effort needed to produce a reward. There was thus a high percentage of both reverse-contingency and prepotent neurons, and both types were widely distributed, as were the many delay and effort neurons. We conclude that inhibitory control derives from many sources in the telencephalon, with subcortical structures such as the striatum and the amygdala being especially enriched in reverse contingency neurons, and cortical structures such as the orbital prefrontal cortex being enriched in prepotent neurons.

Considerable progress has been made on this project during the present reporting period. Subjects have been trained on both the valuation of symbols and the learning of symbol-action associations. Two subjects have learned a series of association between value and a symbolic stimulus (positive or neutral). We have also mastered the methods required to monitor outputs of the autonomic nervous system in these subjects. In the next year, we will determine whether and how the amygdala contributes to the neural valuation signals for both familiar symbols and novel ones. We will also study amygdalo-frontal interactions for both active choices (based on instrumental learning) and passive learning (based on Pavlovian conditioning). For a book on The Neurobiology of Sensation and Reward (J. A. Gottfried, editor), we have also produced a novel and comprehensive review of the way in which the brains of different species deal with reward and other aspects of valuation. We proposed that through top-down, biased competition, mammals can take advantage of parallel memory systems, in which contradictory experiences lead to competing memories. This knowledge allows mammals to explore and exploit a changeable environment, with sufficient behavioral flexibility to permit different systems to control behavior under different contextual circumstances. As for humans, we retain many of the traits of vertebrates and other mammals but, in addition, our ancestors evolved the granular prefrontal cortex and high-order sensory areas such as the inferotemporal visual cortex. The former permits the use of reward-specific sensory information that has been dissociated from its emotional, motivational, and affective attributes, as well as the attachment of value to rules, strategies and other cognitive constructs. This conceptual understanding will inform the experimental results obtained on this project. In the other aspect of this project, one subject has been trained to learn symbol-action associations. The novel aspect of this project involves the recording and analysis of neuronal avalanches, which are spatiotemporal patterns of synchronized activity that occur spontaneously in superficial layers of the mammalian cortex under various experimental conditions. Discovered by our collaborator Dietmar Plenz and his colleagues, this structured form of spontaneous activity reflects the rapid propagation of locally synchronized activity, which recurs at intervals of several hours. The behavior of neuronal avalanches is typical of physical systems at a critical point for phase transitions. Our hypothesis predicts that, compared to avalanche dynamics for well-learned associations between symbols and actions, reversals in these associations will trigger an increase in neuronal avalanches, as will rapid learning. The activity of prefrontal cortex and premotor cortex will be compared to test the hypothesis that the former is more important early in learning symbol-action associations, whereas during consolidation the latter plays the more important role.

Considerable progress was made on this project during the past reporting year. In a key aspect of this project, we studied neuronal activity as subjects discriminated relative durations and distances. Our analysis focused on the frontal cortex. In a relative duration task, subjects had to report whether a red stimulus or a blue stimulus had lasted longer, when these two stimuli could appear in either order. During a delay period between the two stimuli, many neurons reflected the absolute duration of the first stimulus and others encoded which stimulus, red or blue, had appeared first. During a delay period after the second stimulus, the largest population of neurons encoded whether the first or second stimulus had lasted longer, and a substantial population encoded which stimulus, red or blue, had lasted longer. In a relative distance task, an overlapping population of neurons encoded whether the first or second stimulus had appeared farther from the center of a video screen and/or whether the red or blue stimulus did so. These results showed that frontal cortex neurons encode relative durations and/or distances, including both the stimuli (red versus blue) and the order that they appeared (first versus second). In another key aspect of this project, we completed the first study of the frontal pole cortex. The frontal pole cortex is the part of prefrontal cortex that expands most dramatically during primate evolution, and we studied its neuronal activity. In addition, we contrasted the properties of neurons in the frontal pole cortex with those of cells in the dorsolateral and orbital prefrontal cortex. The task we used, a strategy task, required that subject's remember their previous goal (left or right) until a cue appeared, which instructed one of two strategies: stay or shift. Stay cues required a saccadic eye movement to the previous goal;shift cues required a saccade to the other goal. One block of trials had visual strategy cues (vertical versus horizontal bars or yellow versus purple squares), whereas in a separate block of trials the strategy cue was one versus two drops of fluid. Feedback followed each saccade. In the visually cued strategy task, many frontal pole neurons encoded whether the monkeys had chosen the left or right goal, and they did so only at one time during the trial: just before and/or after feedback. Importantly, they did so for both correct and incorrect choices. Frontal pole cells did not encode the anticipation or delivery of fluid when it served as a cue, nor did they encode the visual cues, strategies, memory for previous goals, future goal choices or plans for future movements. Thus, during a brief period near feedback, the frontal pole cortex encodes which goal the subjects selected, a signal that lasts longer when they need to remember their own choices. Our findings indicated that the frontal pole cortex plays a role in monitoring self-generated choices, which could account for its dramatic expansion during human evolution. This project also involved the exploration of neuronal mechanisms involved in selecting future goals. In this part of the project, we studied the choice of a future goal based on a strategy task similar to the one described above. In a previous reporting year (Fiscal Year 2007), we found that one population of prefrontal neurons encodes future goals (F cells) and a nearly completely separate population encodes previous goals (P cells). We found that the prefrontal cortex encodes these goals and strategies whether the subject chooses a goal based on the correct or incorrect strategy. This project came to a conclusion during Fiscal Year 2009.

Considerable progress was made on this project in the past reporting year. The first part of this project involved the role of the motor parts of the frontal lobe in skill learning, particularly the transfer of skills learned with one hand to the other hand. In collaboration with Leonardo Cohen and his colleagues in NINDS, we found with functional brain imaging that the supplementary motor area (SMA) and its principal thalamic nucleus had more activity when a skill transfered well compared to when it transfered poorly. Furthermore, using repetitive transcranial magnetic stimulation during skill learning, we found that disrupting neural processing in the SMA blocked such transfer. These findings, reported in the journal Current Biology, provided direct evidence for an SMA-based mechanism that supports intermanual transfer of motor skill learning. Further work published in the Journal of Neuroscience studied the timing of this contribution relative to movement. In this aspect of the project, people learned a 12-item sequence, practiced with their right hand. The SMA's contribution to the transfer of this skill was blocked only when repetitive transcranial magnetic stimulation occurred during the period between movements, when the memory of a prior movement contributed to the encoding of specific sequences. These results provided insight into frontal lobe contributions to procedural knowledge. In collaboration with a psychologist from the University of Virginia, Daniel Willingham, we also reviewed the field's current understanding about the learning of such sequential-movement skills. In an article for the New Encyclopedia of Neuroscience, we explained that the motor system learns how the body interacts with the world and uses this knowledge to plan movements and produce the forces needed to reach targets. It does so, in part, by learning to correct both previous and ongoing errors. The motor parts of the frontal lobe, in concert with the cerebellum, correct errors made on previous movements, and they act in concert with the basal ganglia to correct ongoing movements. We also made progress in understanding the development of the frontal lobe, both in relation to other cortical areas and in relation to its evolutionary history. In collaboration with Judith Rapoport, W. Philip Shaw and their colleagues in the Child Psychiatry Branch of NIMH, we described regional development of cortical thickness. We found that the regional patterning of cortical growth aligned with aspects of architectonic maps, putting these brain maps in a novel, developmental perspective. Polysensory and high-order areas of cortex, which are the most complex areas in terms of their laminar structure, had the most complex developmental patterns, as well. Structurally less complex cortical regions, including many limbic areas, showed simpler growth patterns. From a comparative perspective, many of the areas with relatively simple developmental patterns have clearly identified homologues in all mammalian brains and thus likely evolved in early mammals. By contrast, all of the regions innovated or dramatically expanded in primates, such as the granular prefrontal cortex and high-order sensory areas such as the inferotemporal cortex, have complex developmental trajectories. In collaboration with a neuroimager at the University of California (Los Angeles), Russel Poldrak, we wrote a second article for the New Encyclopedia of Neuroscience. This review advanced the idea that the basal ganglia resolves the selection demands that confront behaving organisms. This function can be viewed as an aspect of error reduction, as described above for basal ganglia, and it requires correctly prioritizing, scheduling, planning, and sequencing behaviors, as well as controlling the way in which context elicits the correct behavior for a given circumstance. Finally, work on this project included a new hypothesis about the function of the prefrontal areas, in the context of its evolutionary history, published in Trends in Neuroscience (TINS). Contrary to some views of the frontal lobe, the largest part of the primate prefrontal cortex has no homologue in other mammals. The TINS article proposed that the newly evolved prefrontal cortex encodes, represents and stores knowledge about behaviors, including its consequences. This project came to a conclusion during Fiscal Year 2009.