Cortical plasticity underlying tactile stimulus learning

1. Adult monkeys were trained in a tactile two-bar recognition task. Bar A delivered sinusoidal vibrations to the distal phalanx of finger II-IV and bar B to the middle, proximal, and partly to the distal phalanx of finger III. Each stimulus train was composed of a sequence of brief vibrations that were delivered alternatively by bar A and B. The monkeys’ task was to recognise two successive sinusoids delivered to one of the bars at the end of a stimulus train. Extensive training led to progressive improvements in the task in all monkeys. 2. The topographic organisation of the hand surfaces was determined in the cortical area 3b in the hemisphere contra-lateral to the trained hand using electro-physiological recording techniques. The representation of the trained hand was compared to the representation of the untrained hand in the ipsi-lateral hemisphere (intra-subject control). 3. The behavioural training selectively altered the cortical representations of the skin surfaces involved in training. The training-related neuronal reorganisation was not observed in the cortical representation of the untrained hand. 4. The reorganised hand representations contained zones where inputs from synchronously stimulated skin sites were represented together (input integration). These novel representations of the input domains of the trained bars were coexistent with single segment representations of trained and untrained skin sites (conservative skin field representation). In each monkey the bar-specific representational zones were unequal in size. The dominant representational zone matched the behavioural preference of the monkey defined from psychophysical data. 5. The localisation of the representation of the input domain of bar A at the rostral border of area 3b and the localisation of the representation of the input domain of bar B within the caudo-medial region of the finger map topographically separated both zones from each other (input segregation). Input segregation was also provided by the reorganisation of dorsal inputs. 6. Neuronal receptive fields in the experimental hemispheres were significantly enlarged compared with receptive fields in control hemispheres. Receptive field overlap was substantially increased in the bar-specific input domains with sharp discontinuities between the overlap zones.