Journal of Experimental Psychology: Human Perception and Performance 1997, Vol. 23, No. 6,1708-1726 Copyright 1997 by the American Psychological Association, Inc. 0096-1523/97/13.00 Neuronal Correlates of Sensorimotor Association in Stimulus-Response Compatibility Alexa Riehle National Center for Scientific Research Sylvan Kornblum University of Michigan Jean Requin National Center for Scientific Research Neuronal mechanisms underlying stimulus-response (S-R) associations in S-R compatibility tasks were identified in 2 experiments with monkeys. Visual stimuli were presented on the left and right calling for left-right movements under congruent and incongruent S-R mapping instructions. High- and low-pitched tones calling for left-right movements were presented to the left and right ear, and the stimulus side was irrelevant. Single neurons sensitive to the S-R mapping rule were found in the primary motor cortex. The large overlap between the neuronal populations sensitive to the stimulus side, the S-R mapping rule, and the response side, respectively, is consistent with the idea that sensory-to-motor transformation is a continuous rather than a discrete process. Results partly support the hypothesis that the increase in reaction time with incongruent mapping is caused by the automatic activation of the congruent, but erroneous, response. The study of neuronal mechanisms underlying the associa- tion of sensory inputs and motor outputs is key to understand- ing the nature and acquisition of skills. A behavioral paradigm that seems ideally suited to address this issue is stimulus-response compatibility (SRC), so named by Paul Fitts, who first reported it (Fitts & Deininger, 1954; Fitts & Seeger, 1953). Here, stimuli and responses are paired in different ways to construct tasks in which performance (i.e., reaction time [RT] and accuracy) is determined not by stimulus or response factors alone but by their interactions (i.e., their association). For example, if the stimuli consist of two lights, one on the left and the other on the right, and the responses consist of two keys, one on the left and the other on the right, performance is better if the responses are made with ipsilateral keys than with contralateral keys. This is true regardless of whether the responding limbs themselves are ipsilateral or contralateral, thus eliminating anatomical fac- tors as the basic cause. SRC effects are among the most robust findings in the behavioral experimental literature (for recent reviews, see Kornblum, 1992; Proctor & Reeve, 1990), and reflect central cognitive associative processes. Alexa Riehle and Jean Requin, Center for Research in Cognitive Neuroscience, National Center for Scientific Research, Marseille, France; Sylvan Kornblum, Mental Health Research Institute, University of Michigan. Jean Requin died on June 21, 1996. This work was supported partly by Air Force Office of Scientific Research Grant F49620-94-1-0020 and North Atlantic Treaty Or- ganization Grant 860125. We thank Michele Coulmance for computer programs and Nicole Vitton for help in collecting the data. Correspondence concerning this article should be addressed to Alexa Riehle, Center for Research in Cognitive Neuroscience, National Center for Scientific Research, 31, chemin Joseph Aiguier, 13402 Marseille Cedex 20, France. Electronic mail may be sent via Internet to ariehle@lnf.cnrs-mrs.fr. Kornblum, Hasbroucq, and Osman (1990) proposed a general model of SRC effects. According to the model, the dimensional overlap (DO) model, if the dimensions or attributes of a stimulus set are perceptually, structurally, or conceptually similar to those of the response set (i.e., if the dimensions overlap), then the presentation of a stimulus element automatically activates its corresponding response element. If the mapping instructions define this automati- cally activated response as correct (congruent mapping), then it is executed quickly; if the mapping instructions define a different response as correct (incongruent mapping), then the automatically activated response is first aborted, the correct response is identified, and it is then executed. Both the abort and the response identification process in the latter cases add to the time required for the execution of the correct response. Automatic activation of a response is not restricted to the case in which the overlap is between the relevant stimulus dimension and the response. It also occurs if the irrelevant stimulus dimension overlaps with the response. One of the earliest such examples is from a study by Simon and Small (1969), who instructed subjects to press a left key to a high-pitched tone and a right key to a low-pitched tone. The tones were randomly presented to the left and the right ear. Although the stimulated ear was irrelevant to the perfor- mance of the task (i.e., its correlation with the response was zero and subjects were instructed to ignore it), RT was faster when the side of the response key and the side of the stimulated ear corresponded than when they did not corre- spond. Although SRC effects tap into the nature of the linkages between sensory input and motor output, few attempts have been made to study the neural basis of these effects. Some evidence supporting the DO model's automatic response activation hypothesis has been obtained from psychophysi- 1708