* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Representations in the Human Prefrontal Cortex
Embodied language processing wikipedia , lookup
Brain morphometry wikipedia , lookup
Clinical neurochemistry wikipedia , lookup
Haemodynamic response wikipedia , lookup
Human multitasking wikipedia , lookup
Neuromarketing wikipedia , lookup
Functional magnetic resonance imaging wikipedia , lookup
Activity-dependent plasticity wikipedia , lookup
Environmental enrichment wikipedia , lookup
Limbic system wikipedia , lookup
Cortical cooling wikipedia , lookup
Neurogenomics wikipedia , lookup
Neuroscience and intelligence wikipedia , lookup
Artificial general intelligence wikipedia , lookup
Optogenetics wikipedia , lookup
Affective neuroscience wikipedia , lookup
Neurolinguistics wikipedia , lookup
Neuroinformatics wikipedia , lookup
Emotional lateralization wikipedia , lookup
Brain Rules wikipedia , lookup
Time perception wikipedia , lookup
History of neuroimaging wikipedia , lookup
Neuroplasticity wikipedia , lookup
Feature detection (nervous system) wikipedia , lookup
Evolution of human intelligence wikipedia , lookup
Neuroanatomy wikipedia , lookup
Nervous system network models wikipedia , lookup
Human brain wikipedia , lookup
Cognitive neuroscience of music wikipedia , lookup
Neuroanatomy of memory wikipedia , lookup
Neuropsychology wikipedia , lookup
Synaptic gating wikipedia , lookup
Neuroesthetics wikipedia , lookup
Embodied cognitive science wikipedia , lookup
Neural correlates of consciousness wikipedia , lookup
Neurophilosophy wikipedia , lookup
Neuropsychopharmacology wikipedia , lookup
Holonomic brain theory wikipedia , lookup
Inferior temporal gyrus wikipedia , lookup
Aging brain wikipedia , lookup
Cognitive neuroscience wikipedia , lookup
Metastability in the brain wikipedia , lookup
Prefrontal cortex wikipedia , lookup
CU R RE N T D I R E CT I O NS IN P SYC H OL OGI C AL SC I EN C E Representations in the Human Prefrontal Cortex Edward D. Huey, Frank Krueger, and Jordan Grafman Cognitive Neuroscience Section, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland ABSTRACT—The prefrontal cortex (PFC) in humans has been studied for more than a century, but many crucial questions about its functions remain unanswered. This paper will highlight a few key differences between human and animal PFCs, and between the human PFC (HPFC) and other parts of the human brain. We then make a case that the HPFC is critically important for executing behaviors over time and integrating disparate information from throughout the brain. Finally, we will focus on our position in the current debate regarding how the HPFC performs its functions and discuss future directions for research. KEYWORDS—human prefrontal cortex; frontal lobes In humans, the prefrontal cortex (PFC) occupies approximately one third of the entire cerebral cortex, consisting of the area anterior to the supplemental motor area and the premotor cortex (see Fig. 1). The human PFC (HPFC) is central to many of the behaviors that make us ‘‘human,’’ including language, reasoning, decision making, social interactions, planning, and creativity. This review will discuss some of the differences between human and animal PFCs and between the HPFC and other areas of the human brain. We will then review a current debate about how the HPFC functions in light of its unique characteristics. We conclude with a discussion of some suggested future directions. UNIQUE CHARACTERISTICS OF THE HPFC COMPARED TO THE PFC IN OTHER ANIMALS The neurobiological basis of the unique intelligence of humans is complex (see Roth & Dicke, 2005, for a complete review). The intelligence of a species is associated with its overall brain size, the size of its PFC relative to the rest of its brain, the number of Address correspondence to Jordan Grafman, Cognitive Neuroscience Section, National Institute of Neurological Disorders and Stroke, Bldg. 10; Room 5C205; MSC 1440, National Institutes of Health, Bethesda, Maryland 20892-1440; e-mail: [email protected]. Volume 15—Number 4 neurons in its brain, and its neuronal interconnectivity (Roth & Dicke, 2005). Humans do not have the largest brains in the animal kingdom (some larger mammals, such as elephants and whales, have larger brains). Humans have one of the largest relative PFC sizes in the animal kingdom, but other animals (for example, some species of whale) arguably have a larger relative PFC. The human brain appears to be unsurpassed in overall number of neurons (although the African elephant is very close) and in its degree of neuronal interconnectivity, especially in the PFC (Elston, 2003). Likely, human intelligence is related to a combination of large brain and relative PFC size and high neuron number, density, and interconnectedness (Roth & Dicke, 2005). UNIQUE ATTRIBUTES OF THE HPFC COMPARED TO OTHER AREAS OF THE BRAIN The HPFC evolved recently and rapidly compared to other areas of the brain. Within a few million years, the human brain has tripled in size, with the largest proportion of that increase occurring in the frontal regions—especially Brodmann’s area 10 (part of the anterior PFC including the frontal pole; Roth & Dicke, 2005). This increase in size and complexity of the HPFC is associated with tool use, art, language, culture, consciousness, and other uniquely human abilities. The PFC likely evolved from adjacent posterior structures such as the premotor cortex and the supplemental motor area. The HPFC also has a high degree of interconnectivity. This applies both to individual HPFC neurons as compared to neurons from other areas of the human brain (Elston, 2003) and to the connectivity of the systems and structures of the HPFC with each other and with more posterior brain structures (Wood & Grafman, 2003). The HPFC’s high interconnectivity on cellular and structural levels likely contributes to its ability to integrate input from many sources in order to implement more abstract behaviors. The PFC is uniquely oriented to time. Almost 70 years ago, Jacobson made the important observation that monkeys with damaged PFCs had difficulty remembering which container held Journal compilation r 2006 Association for Psychological Science. No claim to original U.S. government works. 167 Representations in the Human Prefrontal Cortex Fig. 1. Anatomy of the human prefrontal cortex (HPFC). The HPFC can be divided into the anterior PFC (APFC), dorsolateral PFC (DLPFC), ventrolateral PFC (VLPFC), and medial PFC (MPFC). From ‘‘Prefrontal and Medial Temporal Lobe Interactions in Long-Term Memory,’’ by J.S. Simons and H.J. Spiers, 2003, Nature Reviews Neuroscience, 4, p. 638. Copyright 2003, Macmillan Magazines Ltd. Reproduced with permission. Figure derived from Neuroanatomy: Text and Atlas (2nd Ed.), by John H. Martin, 1996, Stamford, CT: Appleton & Lange, pp. 458–459. Copyright 1996, McGraw-Hill companies. Used with permission. food if there was a delay between observing the researcher place food in a container and choosing a container to open (Jacobsen & Nissen, 1937). Later researchers discovered a unique property of neurons in the PFC of monkeys that related to this early finding: The neurons can continuously fire during an interval between a stimulus and a delayed judgment about the stimulus. Neurons in other brain areas of monkeys are directly linked to the presentation of a single stimulus, and if they demonstrate continuous firing, it is probable that they are driven by neurons in the PFC or by continuous presentation of the stimulus. In functional magnetic resonance imaging (fMRI) studies, humans also demonstrate continuous PFC activation between a stimulus and a delayed judgment about the stimulus (Curtis & D’Esposito, 2003). If the firing of neurons in the PFC is linked to activity that ‘‘moves’’ the subject toward a goal rather than reacting to the appearance of a single stimulus, then potentially those neurons could continuously fire across many stimuli or events until the goal is achieved or the behavior of the subject is disrupted. This observation of sustained firing of PFC neurons across time and events has led many investigators to suggest that the HPFC must be involved in maintaining a stimulus across time. CURRENT APPROACHES: PROCESS VERSUS REPRESENTATION A key debate in current research on the HPFC is whether a process or a representational view best explains its function. Traditionally, HPFC function has predominantly been studied with a processing approach. Such an approach takes the view that cognition in the PFC can primarily be described in terms of cognitive processes independent of the material (representations) being processed. In this view, PFC processes such as 168 switching, maintenance, and inhibitory control are computational procedures or algorithms operating upon knowledge stored in other, posterior parts of the brain. The representational view, in contrast, focuses on unique kinds of knowledge hypothesized to be stored as memories in the HPFC. A representation can be strengthened by repeated exposure to the same or similar knowledge element and is a member of a psychological and neural network in the HPFC composed of multiple similar representations. One or more representations can remain activated over a period of time and compete with activation of other sets of representations by facilitation or inhibition of neural activity. In this view, a process such as ‘‘inhibition’’ can be reinterpreted as the activation of knowledge stored in the prefrontal cortex that enforces long-term goals based on prior experience and suppresses activation in those brain regions concerned with rapid responses that may be inappropriate to achieving those goals. Models of HPFC function can take a primarily processing or representational approach, or can take a hybrid of the two approaches. An example of a processing model is the adaptivecoding model. This model proposes that HPFC neurons are substantially adaptable or programmable to meet current behavioral demands (Duncan & Miller, 2002). It emphasizes global attention—the selective focusing on relevant stimuli and the role of the HPFC in directing the activity of other brain systems. In support of this model, a high proportion of PFC neurons show activity in monkeys performing disparate tasks. The adaptivecoding model is an example of a processing model because it emphasizes the lack of regional specialization in the PFC and instead posits a nonspecific general processing role. An example of a hybrid model is the temporal-organization model (Fuster, 2002). This model emphasizes the role of the PFC Volume 15—Number 4 Edward D. Huey, Frank Krueger, and Jordan Grafman in the temporal organization of speech and behavior. It distinguishes between PFC representations and processing but asserts that the PFC is both a permanent memory store and the site of processes such as working memory, attention, monitoring, and planning. (See Wood & Grafman, 2003, for further discussion and comparison of different models of PFC function.) As an example of a representational model, we will discuss a particular model proposed by our laboratory. Our group has suggested that the HPFC stores a unique type of knowledge in the form of structured event complexes (SECs). SECs are representations composed of goal-oriented sequences of events that are involved in planning and monitoring complex behavior. Aspects of SECs are represented independently in the HPFC but are encoded and retrieved as a single episode. The stored characteristics of the SEC representations (e.g., frequency of exposure) form the bases for the strength of representation in memory (e.g., higher frequency of exposure leads to stronger representations) and the relationships between SEC representations (e.g., high-frequency representations are likely to inhibit low-frequency representations). According to this theory, the representations in the HPFC differ from the type of representational memory people are more used to— e.g., semantic memories contained in the inferolateral temporal lobe. Semantic memories (e.g., the memory that allows one to name the capital of France) are explicit (associated with conscious awareness) and declarative (consciously recalled), but can be implicitly primed. Memories in the HPFC are established by abstracting information across time and events that integrates elements of knowledge (e.g., themes that cannot be further reduced to their simpler elements without losing their meaning). The representations are usually implicitly executed (especially since they are activated over long periods of time even in the absence of directly relevant stimuli) and are typically goal related. However, we can also explicitly access these memories via translational and time-compressed mechanisms (e.g., describing a plan by listing events, creating a cartoon, or in modern times, creating a video). In our opinion, the processing approach to PFC function is a fundamental shift away from how cognitive neuroscientists understand other types of brain function. Processing models suggest that the PFC is minimally committed to long-term storage of knowledge, as compared to posterior cortical regions in the temporal, parietal, and occipital lobes. We argue that representational models of the HPFC may be superior to process models for theoretical and empirical reasons: Representational models are more consistent with evolutionary theory, they make testable predictions, and they too can be supported by the experimental evidence. First, much of the cortex other than the PFC is organized on a representational model, with more posterior parts of the cortex processing simpler information and more anterior parts of the cortex processing more complex information. For example, in the visual system, cells in the primary visual cortex are activated by simple patterns, whereas cells in more anterior visualprocessing areas of the brain respond to more complex, inte- Volume 15—Number 4 grated patterns. It seems likely that the development of the PFC would follow the same representational model as the more posterior cortex from which it was derived, especially given the rapid increase in PFC size and complexity that occurred with the evolution of modern humans. A parsimonious hypothesis is that the PFC encodes unique representations of complex behaviors (e.g., social inferences) in a manner similar to the posterior cortex. According to this hypothesis, the HPFC would contain integrative representations more complex than those contained in more posterior brain areas, and yet would access representations contained in posterior areas, similar to the way visual stimuli are processed. For example, an HPFC encoding of a plan that takes place in an office could integrate the goal-directed use of a telephone, but the memory of the appearance and simple function of telephones would be contained in more posterior structures. Representational stores in different locations in the PFC (and elsewhere) would then become bound together as part of an episode, with the involvement of memory structures such as the hippocampus (O’Reilly & Rudy, 2000). As discussed earlier, the properties of the HPFC—that it integrates and maintains information about stimuli over time—make it uniquely suited to store representations of episodes that unfold temporally. Second, a representational approach forces investigators to define the nature and location of memories in the HPFC, and this is well suited for generating testable hypotheses. Processing models less often suggest specific hypotheses or brain regions involved because, by definition, the processes are independent of the specific stimuli being processed. We also have wondered why such a proportionally large brain region as the PFC would be the repository of so few ‘‘processes.’’ Third, there is growing empirical data supporting representational models. SECs appear to be selectively processed by anterior PFC regions. Errors in event sequencing can occur with preservation of aspects of event knowledge. Thematic knowledge can be impaired even though single-event knowledge is preserved. Additionally, SEC frequency can affect the ease of retrieval of SEC knowledge (Grafman, in press). REGIONAL PREDICTIONS Because representations are stored in specific brain areas, we can use our SEC model to make predictions, based on current data, about the regions of the PFC associated with different aspects of specific PFC functions such as planning. For this purpose, we can crudely divide the prefrontal cortex into left and right, medial and lateral, dorsal and ventral, and anterior and posterior sectors. There is evidence that the left PFC focuses on the specific features of individual events (including features and meanings) that make up a plan whereas the right PFC mediates the integration of information across events (including making sense of the plan as a whole and features at the macro-plan level such as themes and morals; Grafman, Spector, & Rattermann, 2005). We hypothesize that the medial PFC stores key features 169 Representations in the Human Prefrontal Cortex of predictable overlearned cognitive plans that have a contingent relationship with sensorimotor processes and are rarely modified. The lateral PFC would store key features of plans that are frequently modified to adapt to special circumstances. The ventral PFC is concerned with social-category-specific plans, which often have an emotional component (e.g., how to ask a member of the opposite sex for a date). The dorsal PFC is concerned more with aspects of plans representing mechanistic activities without a social component (e.g., repairing a food processor). Finally, the anterior PFC tends to represent plans of long duration composed of many events, whereas the posterior PFC tends to represent plans and actions of short duration and fewer events (e.g., a simple overlearned association). Since no single PFC region would represent all features or components of a plan, specific plans tend to evoke selected patterns of PFC activation. Any region could participate in plan processing depending on the type of plan, with the different plan (and cortical) subcomponents being differentially weighted in importance (and activation) depending on the kind of plan, the moment-by-moment demands of the plan, and previous experience with the plan. For example, the left anterior ventromedial PFC would be expected to represent a long, multi-event sequence of social interactions (i.e., a social plan) with specialized processing of single component events including the computation of their temporal and sequential dependencies and primary meanings (Grafman et al., 2005). We have begun investigating these proposed differential contributions of PFC subregions. Positive evidence has been found for the representation of several different SEC components within the HPFC. For example, one of us (Frank Krueger) performed an fMRI study in humans that detected that highfrequency SECs are encoded in the posterior medial Brodmann’s area 10 and low-frequency SECs are encoded in the anterior medial Brodmann’s area 10. There is little in the way of negative studies of this framework, but many predictions of the SEC framework have not been fully explored to date and could eventually be falsified. testable hypotheses. For example, a general prediction would be that patients with HPFC damage should show selective deficits in performance in novel situations for which they cannot access representations to aid them. Furthermore, the pattern of deficits should show regional specificity depending on the location of the lesion in the HPFC. For example, a specific prediction would be that damage to the left anterior ventromedial PFC would impair access to long multi-event sequences of social information and cause special difficulty processing the meaning and features of single events within such sequences, including difficulty computing their sequential dependencies. The development of normed and validated research tools to investigate tasks such as social interactions performed by the HPFC will be required to better investigate the neuropsychology of this brain area. Experiments using virtual-reality technology have the potential to delineate the functions of the HPFC in a controlled yet more naturalistic manner than conventional neuropsychological testing. Innovative imaging technologies have great potential; for example, simultaneous fMRI scanning of multiple subjects (termed ‘‘hyperscanning’’) allows real-time imaging of the neural correlates of social interactions. The elucidation of genetic influences on the HPFC is also an important research topic. Because of the role of the HPFC in storing and executing social behavior, findings on the HPFC have important ethical and societal implications. For example, if a set of clearly articulated and understood social behaviors are found to be additively and significantly associated with particular imaging findings, such findings could be used by the legal system or potential employers to evaluate and screen for those behaviors. And because of its standing in the hierarchy of cognitive abilities, the HPFC is an ideal target for interdisciplinary investigations by, for example, economists, political scientists, historians, and educators. Although the HPFC has been researched for more than a century, there remain many important unanswered and socially relevant questions about the functions of this, the most ‘‘human’’ of brain areas. SUMMARY Recommended Reading Miller, E.K., & Cohen, J.D. (2001). An integrative theory of prefrontal cortex function. Annual Review of Neuroscience, 24, 167–202. Moll, J., Zahn, R., de Oliveira-Souza, R., Krueger, F., & Grafman, J. (2005). Opinion: The neural basis of human moral cognition. Nature Reviews Neuroscience, 6, 799–809. Stuss, D., & Knight, R. (Eds.). (2002). Principles of frontal lobe function. Oxford, England: Oxford University Press. Wood, J.N., & Grafman, J. (2003). (See References) The HPFC appears to be important for executing behaviors over time and integrating disparate information from throughout the brain. Although the prevailing view of the HPFC is that of a functional processor region (in contrast to the view of posterior brain structures), we have instead hypothesized that these functions are based on a large set of associative and integrative representations ranging from social attitudes to long-term plans. OUTSTANDING ISSUES/FUTURE DIRECTIONS Further data are required to resolve the process-versus-representation debate over HPFC function. Representational models such as the SEC framework discussed above provide many 170 REFERENCES Curtis, C.E., & D’Esposito, M. (2003). Persistent activity in the prefrontal cortex during working memory. Trends in Cognitive Sciences, 7, 415–423. Volume 15—Number 4 Edward D. Huey, Frank Krueger, and Jordan Grafman Duncan, J., & Miller, E.K. (2002). Cognitive focus through adaptive neural coding in the primate prefrontal cortex. In D. Stuss & R. Knight (Eds.), Principles of frontal lobe function (pp. 278–291). Oxford, England: Oxford University Press. Elston, G.N. (2003). Cortex, cognition and the cell: New insights into the pyramidal neuron and prefrontal function. Cerebral Cortex, 13, 1124–1138. Fuster, J.M. (2002). Physiology of executive functions: The perception– action cycle. In D. Stuss & R. Knight (Eds.), Principles of frontal lobe function (pp. 96–108). Oxford, England: Oxford University Press. Grafman, J. (in press). Executive Functions. In M. Rizzo & R. Parasuraman (Eds.), Neuroergonomics: The brain at work. Oxford, England: Oxford University Press. Grafman, J., Spector, L., & Rattermann, M. (2005). Planning and the Brain. In R. Morris & G. Ward (Eds.), The Cognitive Psy- Volume 15—Number 4 chology of Planning (pp. 181–198). Hove, England: Psychology Press. Jacobsen, C.F., & Nissen, H.W. (1937). Studies of cerebral function in primates: The effects of frontal lobe lesions on the delayed alteration habit in monkeys. Journal of Comparative Physiological Psychology, 23, 101–112. O’Reilly, R.C., & Rudy, J.W. (2000). Computational principles of learning in the neocortex and hippocampus. Hippocampus, 10, 389–397. Roth, G., & Dicke, U. (2005). Evolution of the brain and intelligence. Trends in Cognitive Sciences, 9, 250–257. Simons, J.S., & Spiers, H.J. (2003). Prefrontal and medial temporal lobe interactions in long-term memory. Nature Reviews Neuroscience, 4, 637–648. Wood, J.N., & Grafman, J. (2003). Human prefrontal cortex: Processing and representational perspectives. Nature Reviews Neuroscience, 4, 139–147. 171