We study the activity of populations of neurons in several major subdivisions of the primate frontal lobe in order to better understand its functional organization. We ask how, in mammalian species that have a large number of frontal cortical fields, these areas act together as well as separately in the selection and control of behavior. The primate frontal lobe consists of two major subdivisions: the prefrontal cortex (PF) and the precentral motor areas, the latter of which include the premotor cortex (PM) and the primary motor cortex (M1). The first years of this project were devoted to studying the organization and function of PM, especially in relation to M1. This year, we reviewed this work in relation to cognate fields in Wise et al., Annual Review of Neuroscience 20: 25-42, 1997. More recently, the focus of study has shifted toward critical re-examination of the function of PF, especially in relation to PM. PF has received considerable attention for its role in higher brain function. The prevailing theory for the role of PF is that it subserves working memory, with ventral areas (termed PFv) being involved in remembering object-related information and dorsolateral regions (PFdl) playing a parallel role for spatial information. An alternative theory is that PF is involved in executive functions, a concept which includes generating, evaluating and guiding appropriate actions as well as selecting information relevant to current behavior, in accordance with adaptive behavior-guiding rules. One prediction of the executive function theory is that neuronal activity in PF should reflect behavior-guiding rules in addition to stimulus, response, reward information. In this year's research, a rhesus monkey performed an eye-movement task using two behavior-guiding rules: conditional and spatial, alternating in blocks. With the exception of the color of a fixation point, all events and stimuli (including the stimulus """"""""objects"""""""" presented and their locations) were identical in the two tasks. Our strategy in searching for the """"""""rule"""""""" signals predicted for PF (see Wise et al., Critical Reviews in Neurobiology, 10:317-356, 1996) was analogous to that used in many neuroimaging studies (e.g., PET), in which two images are subtracted. We compared activity in PF during the two rules and interpreted significant differences as evidence of rule dependency. More than half of the PF neurons in the present study showed activity differences that could be attributed to the behavior-guiding rule. Some neurons showed highly selective activity for the spatial rule, some showed selectivity for the conditional rule, and some showed a more complex selectivity involving both rules. These data support the hypothesis that PF plays a role in guidance of behavior according to previously learned rules and that much of the special information processed by PF is managerial in nature. While we did not test and cannot reject the working memory hypothesis on the basis of the present experimental design, recent results indicate that PFv is not needed for object working memory (Rushworth et al., Journal of Neuroscience, 17:4829-4838, 1997), as contemporary doctrine maintains. Our data are consistent with the competing notion that PF has an executive role in managing action through the mediation of behavior-guiding rules (Wise et al., Critical Reviews in Neurobiology, 10:317-356, 1996). Our next step will be to compare rule- dependent activity among different areas of the prefrontal cortex (PFdl vs. PFv vs. the eye movement areas in frontal cortex). Our preliminary analysis indicates a differential contribution from these regions in behavior-guiding rules. We have also tested the hypothesis that the motor areas of frontal cortex shows evolving neuronal activity as the monkeys become more proficient at learning novel motor skills. The hypothesis that neuronal activity levels significantly change in M1 as monkeys adapt to novel visuomotor transforms was confirmed. Approximately half of the neurons showed significant changes in discharge rate. This proportion is roughly similar to that observed in PM and in one of the eye fields of the frontal cortex during conditional motor learning in earlier work on this project. Both forms of motor learning, conditional motor learning and visuomotor adaptation, require the selection and guidance of movements based on visual inputs, but the differences are profound. In the latter case, visuospatial information is thought to be transformed by the neural network in a manner similar to an analogical algorithm, e.g., an angular deviation of some number of degree clockwise or counter clockwise. By contrast, conditional visuomotor learning involves the arbitrary mapping of visual information, which can be spatial and/or nonspatial, onto movements or the targets of those movements. We conclude that in addition to M1, PM also plays a role in learning the novel, analogical visuomotor transforms. It appears that in addition to the role of PM in mediating arbitrary visuomotor mappings, its role in visuomotor adaptation suggests a more general function in visually based selection and guidance of action.
Showing the most recent 10 out of 39 publications
