Given the failure of the ablation method to identify the neural structures/systems that are crucial for cross-modal recognition (CMR) and for tactile discrimination performance (TDP), we injected radioactive 2-deoxy-[14C]glucose (2-DG) into monkeys trained to a high level of CMR or TDP. Nine monkeys were trained to recognize in one sense-modality, to a level greater than 80% correct, the objects experienced in the alternate modality on only a single prior trial. After injection, CMR was continued (Expt. 1a) exactly as before, except that all CMR problems were now in only one direction; or (Expt. 1b) the second--the CMR--trial of each problem was not realized (i.e. no objects were available in the second modality) so that the monkey merely 'expected' the cross-modal trial. Nine other monkeys were trained on a graded roughness discrimination task with conventional 'titration' procedures to a stable level of performance, either (Expt. 2a) without having undergone any cortical removal, or (Expt. 2b) after unilateral removal--contralateral or ipsilateral to the preferred hand--of the posterior insula or of the second somatosensory projection cortex (SII). Exactly the same training procedures were continued in Expt. 2 after injection of 2-DG. Coronal radiographs were made at 490 microns for all 18 monkeys (and also for a 19th, which served as a control in Expt. 1b). The optical densities of the autoradiographs were measured quantitatively in respect of 50 structures or part-structures (e.g. sulci, thalamic nuclei, subcortical structures); they were rated either with the aid of 'pseudo-colours' produced by the computer, or directly by judging the black/white optical densities, in respect of 29 other structures. These 79 structures were then the dependent variables in MANOVAs or ANOVAs, to determine differences between groups; or within groups with respect to the left/right and ipsilateral/contralateral hemispheres; or interaction effects. It was found that certain structures (e.g. the ventral portion of the claustrum, the insula with its extension ventrally into the fundus, nucleus Medialis dorsalis and nucleus Pulvinar oralis) repeatedly, whereas others (e.g. the amygdala) never gave rise to significant outcomes. Structures in the left cerebral hemisphere were frequently found to be more strongly labelled than those in the right hemisphere.(ABSTRACT TRUNCATED AT 400 WORDS)