Noxious stimuli are thought to modify the activity of several populations of central nervous system (CNS) cells, which underlie the sensory-discriminative and motivational-affective aspects of the pain experience on the one hand and generate motor somatic and vegetative reactions on the other (Hardy et al., 1952; Melzack and Casey, 1968; Melzack and Wall, 1988). Neurons responding to noxious input are indeed found both in the spinal cord and at brainstem, diencephalic and telencephalic sites (see reviews in Albe-Fessard et al., 1985; Willis, 1985; Besson and Chaouch, 1987; Price, 1988; Guilbaud et al., 1989; Kenshalo and Willis, 1990). Tissue injury and activation of C-fibers induce a complex temporal and spatial pattern of changes of the discharge frequency, receptive field organization and excitability of central units (Wall, 1989; Dubner, 1991; Woolf, 1991). Plasticity of the nociceptive system is widespread in the animal kingdom and is conceivably important to prevent the injured part of the body from additional trauma (Woolf and Waiters, 1991). So far, most of the available data on nociceptive mechanisms have been obtained by electrophysiological recording in anesthetized or lesioned (eg decerebrate and/or spinalized) preparations. Although this approach has provided invaluable information, it suffers from some limitations. General anesthetics modify neuronal responses to noxious input (Poggio and Mountcastle, 1960; Casey, 1966; Hori et al., 1984; Collins and Pen, 1987; Sandkiihler et al., 1987; Bushnell and Duncan, 1989; Oliveras et al., 1991). Interruption of descending pathways affects the activity of spinal units as well (Willis, 1988; Duggan and Morton, 1988). Under normal circumstances, processing noxious information within the CNS may be modulated by several factors, among which the activity of large diameter afferent fibers (Melzack and Wall, 1965; Wall, 1978), which may be increased by motor reactions elicited by the noxious stimulus and by the behavioral state of the animal (Dubner, 1988). A thorough investigation of the retatio~ p between pain behavior and CNS activity, which may ultimately lead to the alleviation of pain states in humans, therefore requires experiments: performed on unanesthetized, behaving animals. On these grounds, a number of pain models have been~ uced which differ with regard to the kind of stimulus, its duration and site of application and the behavioral measures employed (Vierck and Cooper, 1984; Chapman et al., 1985; Franklin and Abbott. 1989; Dulmer, 1989). Investigators engaged in pain research are committed to the major task of minimizing distress and suffering, eg by keeping to a minimum the intensity of pain and the number of subjects used in the experiments (Zimmermann, 1983).

Electrophysiological studies performed in awake animals may provide a quantitative estimate of the relationship between single units activity, stimulation and behavioral parameters and of the effects of analgesic drugs or manipulations (Casey and Morrow, 1983; Hayes et al., 1981; Oliveras et al., 1986; Kenshalo et al., 1988; Bushnell and Duncan, 1989). A major advantage of the electrophy $ iological approach lies in its excellent temporal resolution (in the order of msec). However, only a few units can be recorded at the same time. A sampling bias exists, related to the characteristics of the recording electrode and to cell size. The location of cells may be reconstructed a posteriori by marking the recording site. However, maps obtained from different penetrations and different animals are subject to some uncertainty, unless neurons are intracellularly stained after recording (Jankowska et al., 1976; Light et al., 1979 …