Sequential motor behavior and the basal ganglia

An important aspect of motor skill lies in the ability to acquire and utilize representations of ordered sequences of motor acts, thereby allowing one to select and execute a sequence of movements as a whole. The importance of this ability becomes evident when one considers alternatives; for instance, having to tie one's shoes every morning based on the conscious selection of each constituent movement of the tying sequence. Understanding the neural control of motor sequences is also important because of the prevalence of human neuropathologies in which movement sequences are impaired differentially [e.g., Parkinson's disease (Benecke et al., 1987)] and because of the widening use of surgical therapies for those disorders (e.g., deep brain stimulation). The importance of basal ganglia (BG) networks for sequential motor control remains a topic of debate (Marsden, 1984; Mink, 1996). Several lines of research suggest that the motor circuit of the BG plays an important role in sequential behaviors (reviewed recently by Graybiel, 2004; Keele et al., 2003). One of the most prevalent hypotheses is that the BG contributes to the “proceduralization” of action sequences (Eichenbaum and Cohen, 2001). Procedural learning provides the capacity for series of events or actions to be recalled as a whole [i.e., as “chunks” (Graybiel, 1998)] with minimal attentional load. It is possible that the BG contributes both to the laying down of new procedural memories (e.g., as a “teacher”) and to the recall or expression of already-learned procedures (Aldridge and Berridge, 1998; Cromwell and Berridge, 1996; Hikosaka et al., 2002a). Several investigators have made a strong case for a role for the BG in switching between components of a familiar sequence (i.e., between the constituent movements of “natural units of action” or “chunks”) (Aldridge et al., 2004; Brotchie et al., 1991; Graybiel, 1998; Keele et al., 2003). Other investigators have challenged the importance of the BG in this function (Exner et al., 2002; Penhune and Doyon, 2002; Rhodes et al., 2004; Smith et al., 2001; Wenger etnal., 1999), especially for the production of short familiar sequences (Verwey et al., 2002). We have explored BG involvement in the production of familiar sequences using a non-human primate version of the serial reaction time task (Nissen and Bullemer, 1987). The SRTT paradigm has been extremely useful for exploring the psychophysics and functional neuroanatomy of human sequential motor behavior (reviewed by Curran, 1998; Keele et al., 2003). In the prototypical SRTT, a subject is prompted to move her/his hand to a seemingly-random series of target locations. The subject can show learning specific to a surreptitiously-ordered portion of the target series in the absence of awareness of its ordered nature. The primary measure of learning is a reduction in reaction times (RTs; the interval from a “go” signal to the onset of movement), in particular as compared with RTs when the subject is presented later with a random target series. Here we describe an SRTT paradigm for monkeys that shows many similarities to what has been described for humans. We also present preliminary results using this task to explore BG encoding of sequencerelated information and the effects on SRTT performance of reversible inactivation of BG output. Our results cast doubt on a major role for the BG motor circuit in selecting constituent movements of sequences that are short and familiar.