Executive frontal functions - Springer Link

9 downloads 0 Views 118KB Size Report
May 10, 2000 - tions: one retrospective, of short-term memory or senso- ry working memory, and the other prospective, of atten- tive set (or motor working ...
Exp Brain Res (2000) 133:66–70 Digital Object Identifier (DOI) 10.1007/s002210000401

REVIEW

Joaquín M. Fuster

Executive frontal functions

Published online: 10 May 2000 © Springer-Verlag 2000

Abstract This chapter presents a conceptual model of the representational and executive functions of the cortex of the frontal lobe derived from empirical evidence obtained principally in the monkey. According to this model, the neuronal networks of the frontal lobe that represent motor or executive memories are probably the same networks that cooperate with other cerebral structures in the temporal organization of behavior. The prefrontal cortex, at the top of the perception-action cycle, plays a critical role in the mediation of contingencies of action across time, an essential aspect of the temporal organization of behavior. That role of cross-temporal mediation is based on the interplay of two short-term cognitive functions: one retrospective, of short-term memory or sensory working memory, and the other prospective, of attentive set (or motor working memory). Both appear represented in the neuronal populations of dorsolateral prefrontal cortex. At least one of the mechanisms for the retention of working memory of either kind seems to be the reentry of excitability through recurrent cortical circuits. With those two complementary and temporally symmetrical cognitive functions of active memory for the sensory past and for the motor future, the prefrontal cortex secures the temporal closure at the top of the perception-action cycle. Key words Prefrontal cortex · Cross-temporal contingencies · Short-term memory · Motor attention · Perception-action cycle

Introduction The entire cortex of the primate’s frontal lobe seems dedicated to organismic action. It can, thus, be considered, as a whole, “motor” or “executive” cortex in the J.M. Fuster (✉) UCLA Neuropsychiatric Institute, 760 Westwood Plaza, Los Angeles, CA 90024, USA e-mail: [email protected] Tel.: +1-310-8250247, Fax: +1-310-8256792

broadest sense. The frontal cortex participates in all aspects of active adaptation of the organism to the environment: skeletal and ocular movement, reasoning, and the spoken language. Visceral actions and emotional behavior are modulated to some extent by orbital and medial areas of the frontal cortex. In this chapter, the focus is on the role of the dorsolateral prefrontal cortex in the cognitive aspects of the temporal organization of action, in other words, in the cognitive aspects of executive function. In the present model, there is no need for a “central executive”. Such a hypothetical entity would be distributed here in dorsolateral prefrontal cortex. It is a reasonable assumption that the cognitive functions of the prefrontal cortex, as those of any other part of the neocortex, consist of the activation and processing within and between distributed networks of neurons representing the environment and the organism within it; in other words, within and between memory networks (review in Fuster 1997b). These networks are organized hierarchically and are highly specific, defined by their connectivity. The memory code, in this sense, is a relational code, and all memory is associative. Memory networks extend across cortical modules and areas, overlapping and extensively interconnected with one another. One neuron or group of neurons anywhere in the cortex can be a part of many memory networks and, therefore, many memories. In the cortex of the frontal lobe, executive networks are hierarchically organized like the perceptual networks of posterior, post-rolandic, cortex. The primary motor cortex is the lowest level of the cortical motor hierarchy; it represents and executes elementary muscular movements. Above it, the premotor cortex is the substrate for more complex movements, defined by goal and trajectory. At the top is the prefrontal cortex. On the basis of neuropsychological data, it appears that this cortical region represents the broad schemas of action in skeletal and speech domains. Furthermore, it seems crucially involved in the executions of those schemas. One of the most consistent components of the prefrontal-lesion syndrome is the difficulty in formulating and enacting plans of behavioral, linguistic, or cognitive action.

67

Temporal organization It follows from these general concepts and assumptions that the cognitive functions of the prefrontal cortex cannot be isolated from those of the rest of the neocortex and the subjacent stages of the executive hierarchy. It is erroneous to localize in the prefrontal cortex cognitive functions that are not localizable. This is true for working memory, for the so-called “central executive,” for spatial memory, and for various forms or aspects of attention. All these cognitive functions are subserved by the frontal lobe, but none of them is localized in it. Only parts of the networks of executive memory are localized there. By neural processing within and between those networks, and between those networks and others elsewhere in the neocortex, the prefrontal cortex plays its role in the temporal organization of behavior, speech, and logical reasoning. An outline follows of physiological processes behind the postulated role of the prefrontal cortex in temporal structures of action. At the root of that role is a set of cognitive operations that mediate cross-temporal contingencies. The dorsolateral prefrontal cortex plays a critical role in that mediation, as is now substantiated in the human and non-human primate. Behaviorally, the best way to study the mediation of cross-temporal contingencies is by use of the delay-task paradigm. In delay tasks, motor acts are contingent on sensory cues that the animal must retain for a period (delay) of seconds or minutes. A single trial in one such task is an example of temporal structure or “gestalt.” Since each trial temporally separates events that are mutually contingent, and since those events are novel – inasmuch as they change randomly between trials – the task is based on the mediation of cross-temporal contingencies. What follows is a brief summary of some of our evidence for: (1) the crucial importance of the dorsolateral prefrontal cortex for the mediation of cross-temporal contingencies; (2) the role of prefrontal networks in active short-term memory, or working memory, the retrospective aspect of cross-temporal contingencies; (3) the role of prefrontal networks in short-term attentive set, the prospective aspect of cross-temporal contingencies; and (4) the mechanisms of short-term active memory. Cooling dorsolateral prefrontal cortex – parts of areas 9 and 46 – in the monkey induces reversible deficits in visual, auditory, and tactile delay tasks (Fuster 1997a). It appears that the animal with functionally depressed prefrontal cortex cannot mediate across time the contingency between a stimulus and the consequent behavioral response to it, especially if the time between the two is long (>5 s). If the task is visual (delayed matching to sample), a similar deficit can be obtained by bilaterally cooling the inferotemporal cortex. There is considerable anatomical and physiological evidence for areal specificity within the dorsolateral prefrontal cortex, with regard to stimuli as well as contingency tasks (Fuster et al. 1982; Petrides et al. 1993a; Petrides and Pandya 1994; Goldman-Rakic 1995). Im-

aging studies corroborate that evidence (Paulesu et al. 1993; Petrides et al. 1993b; Courtney et al. 1996; Owen et al. 1996; Smith et al. 1996; Aguirre et al. 1998; Gabrieli et al. 1998). These studies demonstrate that several prefrontal areas are activated during working memory. We contributed to this evidence with a positron emission-tomography (PET) deoxyglucose in human subjects performing a visual delayed matchingto-sample task (Swartz et al. 1995). The mechanisms of cortical activation in short-term memory can best be investigated in real time at the cellular level. Neuronal phenomena show that the mediation of cross-temporal contingencies depends on two temporally symmetrical functions: (1) active short-term (working) memory and (2) short-term attentive set or motor memory.

Active short-term memory The results from numerous studies clearly indicate that the dorsolateral prefrontal cortex is critically involved in all forms of working memory toward a goal, whether in behavior, reasoning, or speech (Fuster 1997a). Frontal memory in the active state is defined by its teleological quality. All the physiological attributes of single areas and cell groups in the dorsolateral frontal cortex seem to be related to that teleological quality of frontal memory. In that cortex, sensory-cell groupings and memory maps may be simply the pathways of access to executive action. Sensory working memory is one of the cognitive functions at the service of that executive action. It assists in organizing behavior in the temporal domain. Some areas of ventral or orbital cortex receive information related to reward or emotion. That too may have teleological significance, for that information may serve the fulfillment of drives. Those areas have important inputs from the limbic system, notably the amygdala. The involvement of prefrontal neurons in short-term active memory is amply documented by descriptions of single units that respond by activation of firing frequency to the presence of a sensory memorandum for prospective action (e.g., Fuster 1973; Niki 1974; Funahashi et al. 1989). Such units have two characteristics that define them as so-called “memory cells”: (1) specific response to one or more memoranda, and (2) temporal decay of their firing during memorization (delay); this phenomenon is most evident in manual memory tasks with long delays (over 5 s). Those cells are apparently constituents of motor memory networks that are activated by stimuli that need to be retained toward a goal. Their firing frequency is related not only to those stimuli or memoranda, but to the level of correctness of performance of the task. Different sensory memoranda have different prefrontal distributions. In a task with varying probability of association between a visual stimulus and a later manual act, we studied the temporal characteristics of sensory and motor memory cells in dorsolateral prefrontal cortex (Quintana

68

and Fuster 1999). The task is a combination of delayed matching to sample and delayed conditioned discrimination. Each trial begins with the presentation of a color, the initial cue. Following a delay of 12 s, a second visual cue is given, and the animal is required to perform a manual choice that depends on both the first and the second cue; therefore, on the combination of two visual stimuli separated by time. Since each combination of the two stimuli – double contingency – determines whether the response after the delay will be to the left or to the right, the trial-initiating color (red, green, blue, or yellow) predicts with a given degree of probability (100% or 75%) the response direction. In the prefrontal cortex, during performance of that double-contingency task, two broad categories of “memory cells” can be observed, that is, of cells with elevated discharge during the memory period (delay). The first respond with a different level of activation depending on the color of the initial cue; their discharge descends toward baseline level in the course of the delay. They seem to distinguish and remember the colors, and their discharge during that period seems to reflect the decay of working memory. The second type of cells behave in the opposite way. Their firing during the delay is related to the direction of the approaching manual response, before it is prompted by the second cue. These cells appear to anticipate the response. They accelerate their firing as the delay advances. Moreover, the degree of acceleration varies in proportion to the certainty with which the animal can predict the direction of the response. Both types of cells, sensory-coupled and motor-coupled cells, are intermixed, and, therefore, it is not possible to identify a separate cortical topography for each of them. Thus, it seems that, during the delay, there is a transfer of information from sensory-memory cells to motor-set cells. More precisely, there appears to be a temporal transfer from one prefrontal network to another, both connected to the visual networks of posterior cortex (inferotemporal) and the motor networks downstream.

Attentive set The other category of prefrontal units, those that seem to anticipate the action (Quintana and Fuster 1999), may be the substrate for the short-term activation of motor memory, that is, the converse of sensory working memory. Consequently, there seems to be a sensory and a motor short-term memory, one retrospective and the other prospective, the two complementing each other at the service of the frontal executive in the mediation of crosstemporal contingencies. Active prospective memory is also the motor equivalent of sensory attention; it may be called “motor attention.” It would be a function in some way related to, if not the same as, what has been termed the “memory of the future” (Ingvar 1985). The cellular manifestations of motor set acquire special significance in view of the neu-

ropsychological evidence that the dorsolateral prefrontal cortex is profoundly implicated in the formulation and execution of plans of action. The motor-set cells would be the support for the frontal planning functions in the short-term. The sustained activations of prefrontal cells in delay tasks are related to the field potentials observable in the human frontal cortex between mutually contingent and temporally separate events, especially the “contingent negative variation” or CNV. The accelerating cell activations of “set cells” are undoubtedly related to the “Bereitschaftspotential,” or “readiness potential,” which is another negative potential that takes place after the CNV, immediately preceding a pre-instructed motor act. These surface negative potentials, which appear over frontal cortex after a sensory stimulus and before an action that is contingent on it, probably reflect the activation of large neuronal populations involved in working memory and motor set. To summarize, the ramping-up cells and the surface negative field potentials, in particular the “readiness potential,” appear to be electrophysiological manifestations of attentive set, or motor attention, the prospective function of the dorsolateral prefrontal cortex. This would be attention focused on the action in preparation, in other words, focused in the representations of the actions about to take place. It is attention directed to the components of long-term motor memory that are activated successively for the execution of the behavior in progress. Those cells apparently predicting future actions indicate the presence of mechanisms in the dorsolateral prefrontal cortex preparing the motor apparatus for those actions. It is possible that such mechanisms result in the priming of lower motor structures (e.g., premotor cortex, basal ganglia, pyramidal system) for those actions.

The perception-action cycle Finally, to conclude this chapter, a few considerations are in order regarding the mechanisms of functional coordination of working memory and attentive set, the two prefrontal functions mentioned above, in the temporal organization of behavior. What are the mechanisms by which the prefrontal cortex mediates the transfer of information from the past to the future and from the perception to the action? Undoubtedly, these mechanisms involve local transactions, as the two functional types of neighboring cells suggest, and also remote or transcortical processes, as some of our other cellular evidence suggests. Reentry or reverberation through recurrent circuits may play a role in both these kinds of processes, local and remote. Several methods have been used with the purpose of testing the reentry hypothesis. First, an artificial network model was developed with an architecture that was profusely recurrent. We trained that model to “sample” an external input, to retain it in memory, and to elicit

69 Fig. 1 The cortical stages of the perception-action cycle. The arrows represent connections demonstrated anatomically in the non-human primate. (Schematic diagram of the human brain with color-coded representation of the sensoryand motor-processing hierarchies)

an output that was a specified function of the input (Zipser et al. 1993). To train the model, we used the back-propagation method. This is an error-reducing method allowing the network, through multiple iterations, to adjust synaptic weights to keep a stable relationship between input and output, despite fluctuations in the input. The weights stay fixed after the training has been completed, after the network has “learned.” In the fully trained network, a memory trial is simulated by loading an input, that is, the memorandum, and by holding a gate open through a memory period or delay until the recall, when another load signal closes the gate and emits the output. The output cells behave predictably and unremarkably, because they only reflect the input-output function specified by the modeler. Exceptionally remarkable, however, is the discharge of the internal or “hidden units” of the network. With appropriate scaling, these cells appear to behave like real cells in a delay-memory task. It can be concluded from this finding that the firing patterns of cortical cells are consistent with the sustained activation of fully trained recurrent networks with fixed and preestablished synaptic weights. Thus, the model supports the

concept of reentry as a mechanism in the cortical dynamics of short-term memory. The reentry mechanism in short-term memory is also supported by experiments with reversible cortical depression by cooling (Fuster et al. 1985). Monkeys were trained to perform a delayed matching task with colors. After the animals were trained, cooling probes and microelectrode carriers were implanted, bilaterally, on two cortical areas known to participate in the performance of that task, the dorsolateral prefrontal cortex and the inferotemporal cortex. While the animal performed the task, prefrontal or temporal cortex was cooled bilaterally, and units were concomitantly recorded from the noncooled cortex. By cooling either inferotemporal or parietal cortex to 20°C, a reversible impairment is induced in the monkey’s performance of the task. The procedure also modifies the spontaneous and memory – delay period – discharge of the cells in the other (non-cooled) cortex. Under normal conditions, some prefrontal and inferotemporal cells are differentially activated during the memorization of the color sample: they seem to prefer one color in

70

short-term memory over the other colors in the same condition. Under cooling of the other, distant cortex, these cells show lesser differences in their color-related memory discharge; they differentiate less the colors in memory than when that cortex is at normal temperature. Therefore, it appears that, in the absence of input from the remote cortex (under cooling), neurons in both prefrontal and inferotemporal cortex are less active in the memorization of their preferred color. At the same time, the monkey is less able to retain the colors in short-term memory. This leads to the inference that short-term visual memory is maintained active by reentrant, reverberating, neuronal influences between inferotemporal and prefrontal cortices. In broader neurobiological terms, the apparent recurrent excitation between the prefrontal cortex and the sensory association cortex can be viewed as a mechanism of temporal closure at the summit of the perception-action cycle. This cycle is the circular flow of neural processing linking the organism to the environment. Anatomically, the cycle basically consists of two parallel hierarchies of neural structures, one sensory and the other motor, which extend along the nerve axis from the spinal cord to the highest cortex of association (Fig. 1). At all levels, structures are interconnected reciprocally: by feedforward as well as feedback. In new or recently learned behavior, sensory impulses are processed along the sensory hierarchy and into the motor hierarchy. That sensory information is thus translated into action, processed down the motor hierarchy to produce changes in the environment. Such changes elicit new sensory impulses, which are processed in the sensory hierarchy and then modulate further action, and so on. The posterior cortex of association and the dorsolateral prefrontal cortex are part of the processing cycle, if the behavior contains novelty or uncertainty, and has to bridge temporal gaps with short-term memory and attentive set; in other words, if it has to mediate crosstemporal contingencies. As the behavior becomes automatic (e.g., walking, skilled routines), the action is integrated in lower structures (e.g., premotor cortex, basal ganglia), and the processing is shunted at lower levels. This is probably why, with the over-learning of tasks that do not require working memory, metabolic activations disappear from the highest cortical levels of the perception-action cycle (Grafton et al. 1992; Raichle et al. 1994). At the top of the perception-action cycle, the dorsolateral prefrontal cortex integrates actions with perceptions, especially in situations marked by novelty, ambiguity, and complexity. In order to do it, the prefrontal cortex works in close cooperation with other areas of the neocortex and with subcortical structures in lower stages of the executive hierarchy.

References Aguirre GK, Zarahn E, D’Esposito M (1998) Neural components of topographical representation. Proc Natl Acad Sci USA 95:839–846 Courtney SM, Ungerleider LG, Keil K, Haxby JV (1996) Object and spatial visual working memory activate separate neural systems in human cortex. Cereb Cortex 6:39–49 Funahashi S, Bruce CJ, Goldman-Rakic PS (1989) Mnemonic coding of visual space in the monkey’s dorsolateral prefrontal cortex. J Neurophysiology 61:331–349 Fuster JM (1973) Unit activity in prefrontal cortex during delayedresponse performance: neuronal correlates of transient memory. J Neurophysiol 36:61–78 Fuster JM (1997a) The prefrontal cortex anatomy: physiology and neuropsychology of the frontal lobe. Lippincott-Raven, Philadelphia Fuster JM (1997b) Network memory. Trends Neurosci 20:451– 459 Fuster JM, Bauer RH, Jervey JP (1982) Cellular discharge in the dorsolateral pefrontal cortes of the monkey in cognitive tasks. Exp Neurol 77:679–694 Fuster JM, Bauer RH, Jervey JP (1985) Functional interactions between inferotemporal and prefrontal cortex in a cognitive task. Brain Res 330:299–307 Gabrieli JDE, Poldrack RA, Desmond JE (1998) The role of left prefrontal cortex in language and memory. Proc Natl Acad Sci USA 95:906–913 Goldman-Rakic PS (1995) Architecture of the prefrontal and the central executive. Proc Natl Acad Sci USA 769:71–83 Grafton ST, Mazziotta JC, Woods RP, Phelps ME (1992) Human functional anatomy of visually guided finger movements. Brain 115:565–587 Ingvar DH (1985) “Memory of the future”: an essay on the temporal organization of conscious awareness. Human Neurobiol 4:127–136 Niki H (1974) Differential activity of prefrontal units during right and left delayed response trials. Brain Res 70:346–349 Owen AM, Evans AC, Petrides M (1996) Evidence for a twostage model of spatial working memory processing within the lateral frontal cortex: a positron emission tomography study. Cereb Cortex 6:31–38 Paulesu E, Frith CD, Frackowiak RSJ (1993) The neural correlates of the verbal component of working memory. Nature 362: 342–344 Petrides M, Pandya DN (1994) Comparative architectonic analysis of the human and the macaque frontal cortex. In: Boller F, Grafman J (eds) Handbook of neuropsychology. Elsevier, Amsterdam, pp 17–58 Petrides M, Alivisatos B, Evans AC, Meyer E (1993a) Dissociation of human mid-dorsolateral from posterior dorsolateral frontal cortex in memory processing. Proc Natl Acad Sci USA 90:873–877 Petrides M, Alivisatos B, Meyer E, Evans AC (1993b) Functional activation of the human frontal cortex during the performance of verbal working memory tasks. Proc Natl Acad Sci USA 90:878–882 Quintana J, Fuster J (1999) From perception to action: temporal integrative functions of prefrontal and parietal neurons. Cereb Cortex 9:213–221 Raichle ME, Fiez JA, Videen TO, MacLeod A-MK, Pardo JV, Fox PT, Petersen SE (1994) Practice-related changes in human brain functional anatomy during nonmotor learning. Cereb Cortex 4:8–26 Smith EE, Jonides J, Koeppe RA (1996) Dissociating verbal and spatial working memory using PET. Cereb Cortex 6:11–20 Swartz BE, Halgren E, Fuster JM, Simpkins F, Gee M, Mandelkern M (1995) Cortical metabolic activation in humans during a visual memory task. Cereb Cortex 3:205–214 Zipser D, Kehoe B, Littlewort G, Fuster J (1993) A spiking network model of short-term active memory. J Neurosci 13:3406–3420