Ejecución de la pena

The histological reconstruction of 3 monkey brains showed that all cells tested for whole body stimuli presented in different orientations were located in the upper and lower bank of the STS (see Fig. 7.12 and Appendix 4 for one subject). There was no observed clustering of cell types. A chi-square test showed that cells tuned to the upright image are not more frequently found at more anterior sites than at more posterior sites of the STPa (chi^=1.3, df=l, p=0.25, where group A were cells in

response to the upright body stimulus orientation, medium grey area to the horizontal body stimulus orientation, dark grey area to the inverted body stimulus orientation and the black area population response to various control objects. Firing rate is expressed as a percentage of the peak response to the best (in this case the upright) stimulus. Stimulus presentation occurs at point 0 ms. b) Cumulative Response curve of orientation population estimate (10 cells). The colour coding for different stimuli responses is identical to (a). Stimulus presentation occurs at point 0 ms. The average population response rate was calculated after normalising each cell’s activity (S/A=0, Max Response=100). The cumulative population response is defined as the running total of this average population response from 200 ms pre-stimulus onset.

□ UPRIGHT B HORIZONTAL ■ INVERTED ■ CONTROL Threshold S/A 0 500 PO S T -S T IM U L U S TIM E (m s) Q UPRIGHT HORIZONTAL INVERTED CONTROL 0 500 P O S T -S T IM U L U S TIM E (m s)

Figure 7.12. Histological reconstruction. A frontal section of subject E taken at 15 mm anterior to the interaural plane showing the location of cells tuned to upright orientation of the head/body (open circles) and cells generalising across all orientations of the head/body tested (open triangles). The thick lines indicate the brain surface, the thin lines show the boundary between white and grey matter. Cells were located in both the upper bank and fundus of the superior temporal sulcus (see Appendix 4).

sections <15 mm anterior, and group B were cells found in sections 15-17 mm anterior). Cells coding upright heads/bodies were therefore intermixed with cells coding non-upright heads/bodies and cells which generalised across different orientations in the picture plane.

Dis c u s s io n

Given the length of the introductory review of this chapter, it is appropriate to first recap the main findings:

Most human and animal studies indicate that the ventral stream (V4, PIT, AIT and STPa) is involved in the processing of information about an object's appearance including the orientation of the object (see e.g. Dean, 1976; 1991; 1992; 1993; Tanaka, Fujita et al., 1993; Kobatake and Tanaka, 1994). The FT cortex plays an important role in the formation of mental representations of objects for long term memory (Miyashita, 1990; Miyashita et al., 1993). Ideally, for recognition purposes, a single object-centred representation which generalises across viewing conditions would be most efficient. This would allow an object to be recognised irrespective of its perceived view, orientation, size, or even part visibility (i.e. part-occlusions). For behavioural interaction with an object, information about the object's orientation relative to the viewer has to be available to allow and guide interaction. For such visuo-motor control, viewer-centred representations are essential, although information about the objects’ identity is irrelevant. It now appears that such information about 3D orientation (and size) is coded by cells in parietal cortex (area LIP, Sakata personal communication, 1995; Taira et al., 1990; Sakata and Taira, 1994).

Interestingly, different methods of investigation emphasise the importance of different types of representation in the temporal cortex. The study of brain lesioned subjects (both humans and monkeys) indicate the involvement of the temporal cortex in generalising across changes in orientation (see e.g. Wilson, 1957; Cowey and Weiskrantz, 1967; Gross, 1973; Dean and Weiskrantz, 1974; Dean, 1976), whereas lesions in the parietal cortex appear to disrupt identification of an object’s orientation (Pohl, 1973; Ockleford, Milner et al., 1977; Ungerleider and Brody, 1977; Jeannerod et al., 1994). From such lesion work one might assume that viewer-centred information about objects would be found only in the dorsal stream of processing. Single unit

Effect of Rotation 122 recording studies, on the other hand, reveal that the vast majority of cells in the temporal cortex are selective for object features or parts (see chapter V), view (see chapter VI), orientation and/or size (see chapter VIII). These studies contrast with the lesion studies and indicate the predominance of object representations that are highly specific to viewing conditions (i.e. viewer-centred representations). In the same physiological studies, however, a minority of temporal cortex cells are found to generalise across particular viewing conditions such as orientation.

To summarise neurophysiological studies, cells in early visual areas (VI, V2, V4) exhibit orientation-specific coding of elementary features of objects (Heniy et al., 1974; Trotter et al., 1989; Kobatake and Tanaka, 1993). These cells feed into more anterior visual processing areas (PIT, AIT), where cells are selectively responsive to progressively more complex features and object such as faces but still exhibit orientation specific responses (Tanaka et al., 1991; 1992; 1993; Kobatake and Tanaka, 1994). IT cortex in turn projects to cells present in the STPa (Seltzer and Pandya, 1978) which have been shown to be selective for complex objects and which have previously been reported to respond irrespective of the stimulus orientation (Perrett, Rolls et al., 1982; Perrett et al., 1985; 1988). Interestingly, such cells exhibiting orientation-invariant responses to faces appear to require extra time for the processing of unusual orientations of the stimuli (Perrett et al., 1985; 1984; 1988). The empirical work reported here confirms the presence of cells with orientation-invariant responses but also indicates that most coding for complex objects in the STPa is orientation selective as was found in IT shape selective cells. That is, 82% (21/25) of the cells that were selective for the sight of the body tested here also showed some sort of selectivity to a particular orientation of the body. Furthermore, the Orientation Discrimination Index computed for all the cells tested in this study indicates that the cells which do discriminate between different orientation of the whole body stimuli are relatively good at such discrimination.