Selection of actions in the basal ganglia–thalamocortical circuits: review and model

Abstract

The paper reviews the 20-year experience of recording impulse activity of neurons in the basal ganglia–thalamocortical circuits. These recordings were made from patients with Parkinson's disease who failed to respond to conventional medical treatment and who had undergone stereotaxic neurosurgery. When taken together, the results show that: (1) the basal ganglia–thalamocortical circuits become active only when a stimulus is attended or when a movement is voluntarily implemented, i.e. they are involved in the process of selection of an appropriate sensory stimulus for advanced processing and in the process of selection of an appropriate motor action for achieving a certain goal; (2) neuronal circuits responsible for assessment actions and for motor acts are segregated; (3) inhibitory opponent neuronal mechanisms are implemented for initiating and suppressing inappropriate actions; and (4) preparation to make different assessment actions (attentional set) is associated with different preparatory activities. To explain these findings a hypothesis of action programming has been formulated. According to it, the whole of human behavior is divided into separate sensory–motor–cognitive actions, while the brain in turn is divided into separate systems playing different roles in the organization of actions. The system for action selection that includes the basal ganglia–thalamic circuits plays a critical role in initiation of, preparation for, and suppression of these actions. The neuronal mechanisms for the system for action selection including mapping of actions, `winner takes all' operations in the striatum, disinhibition and inhibition processes in the thalamus are suggested and discussed.

Introduction

We live in an overwhelmingly complex world in which many events occur simultaneously and which interacts in many different ways with ourselves, who in turn constantly move, orient to and manipulate objects in this external world. For providing coherent, goal directed behavior, i.e. for coordinating sensory information flow and motor activity and integrating them into a single perceptual-motor act, the brain seems to employ higher-level control operations. In this paper these operations will be referred to as selection processes.

In the sensory domain, the selection process is associated with attention or attentional set, i.e. a preferential processing of a selected source of sensory information in an environment containing multiple channels of information (Moray, 1970; Allport, 1980; Woods, 1990). The selection process in the motor domain is associated with motor preparatory set, i.e. preparing for a selected movement while suppressing the other movements (see review see Evarts et al. 1984). To emphasize the higher control of movement, the term voluntary (vs. reflexive) movement is used (Passingham, 1993). We believe that selection operations are not limited by sensory or motor domains, they are expressed as integral parts of the whole range of human behavior. The ability to select an appropriate action at a given time interval determines cognitive functions of the human brain.

Models of attention proposed over the last 50 years have been based on psychological, neuropsychological and, more recently, on ERP (event-related potentials) and PET (positron-emission tomography) studies. Reviewing all these models is beyond the scope of the paper. However, several insights presented in the literature on attention have had a strong impact on our own views and need to be briefly mentioned here. The most important of these insights are: (1) a notion of cognitive structures (anticipatory schemata) that prepare the perceiver to accept the information selected (Neisser, 1976); (2) a notion of an automatic mismatch operation in the pre-attention mode (Näätänen and Michie, 1979; Näätänen, 1992); (3) a scheme for attentional networks that includes different cortical and subcortical components implementing engage, disengage and shifting operations (Posner and Dehaene, 1994); (4) a parasol metaphor emphasising the active suppression of irrelevant sources of information in the attention mode (Woods, 1990); (5) a notion of the functional system that is constructed in the brain for achieving a certain goal (Anokhin, 1973); (6) the theory of neuronal group selection introduced to describe principles of brain functioning (Edelman, 1987); (7) the searchlight hypothesis articulated to explain the role of the thalamus in attention and consciousness (Skinner and Yingling, 1977; Crick, 1984; Koch and Crick, 1994); and (8) a hypothesis of synchronous large-scale behavior of neural networks proposed to explain how the brain combines different sensory features into a single percept (von der Malsburg and Schneider, 1986; Gray et al., 1990; Engel et al., 1992).

Models explaining the mechanisms of motor control and motor learning have been mainly based on neurophysiological and neuropsychological data. Almost all of them include the basal ganglia–thalamocortical circuits now viewed by a majority of researchers as parallel feed back loops with separate elements in the striatum, the pallidum, the substantia nigra and the thalamus. Although the exact functions of these circuits are still unknown, single unit recordings in these structures indicate that they are involved in preparation for and initiation of internally generated movements (Alexander et al., 1986; DeLong et al., 1992). Several hypotheses on the role of the basal ganglia in organization of behavior have been suggested emphasizing the involvement in different functions, such as in running the sequences of motor programs to complete a motor plan (Mardsen, 1982), in the process of motor learning (procedural memory) (Hikosaka, 1993), in the image forming operation which control the cortical image to drive (Tamai, 1996), in the initiation of movement in a behavioral context-dependent manner (Kimura, 1995), in learning an association between divergent stimuli for the initiation of sequences of movements (Rolls, 1994), in control of polysensory information to the cortex (McKenzie, 1984) and some other hypothetical functions.

During the last two decades in our laboratory at the Institute of the Human Brain in St Petersburg, we have used a unique opportunity for recording impulse activity of neurons and local field potentials from electrodes located in different parts of the basal ganglia, the thalamus and the cortical areas connected to these subcortical structures. The electrodes were implanted for diagnosis and therapy for Parkinsonian patients, patients with obsessive–compulsive disorder and epileptic patients. These people were those informed and consenting patients who remained unresponsive to all conventional forms of treatment and who were recommended by a special commission for stereotactic neurosurgery. The patients voluntarily participated in our investigations, having given written consent.

Several different behavioral tasks were devised to study sensory, motor and cognitive functions. Functional profiles of the evoked responses together with anatomical information were used for identification of electrodes' respective positions. These studies have been published in different Russian journals (Human Physiology, Sensory Systems, Journal of Higher Nervous Activity, I.M. Sechenov Physiological Journal) as well as in different international journals (International Journal of Psychophysiology, Journal of EEG and Clinical Neurophysiology, Psychophysiology). All these results, when put together, reveal a rather lucid and comprehensive picture of the functional features of the basal ganglia–thalamic circuits indicating their involvement in attention and motor preparation. Moreover, a single neurophysiological mechanism for implementing these functions can be proposed.

The goals of this paper are twofold: (1) to review the results of our own recording of impulse activity of neurons in Parkinsonian patients with implanted electrodes and to show the involvement of the subcortical structures in attention and motor preparation; and (2) to suggest a cortico-subcortical mechanism for the selection of actions (including sensory, motor and cognitive acts).