

Flexible computations play a major role in any intelligent system. In the study reported here, the computational flexibility of the human cognitive system was examined in order to better understand the possible alternative flexible computations and the trade-offs in different forms of computational modifications. To explore the computational plasticity of the visual system, twenty-eight subjects were trained to read mirror-reversed letters. After prolonged training subjects acquired a general skill of reading mirror-reversed letters. The subjects were able to read novel mirror-reversed letters at comparable levels of performance to that of reading the mirror-reversed letters used during training. The generality of the skill found in this study is in contrast to Mason (1986), who found that the acquired skill of reading mirror-reversed stimuli is very limited and letter specific. The prolonged learning used in this study evoked a different form of computational modification. Although the subjects in our study acquired the skill of reading mirror-reversed letters, they were unable to read mirror-reversed letters as fast as upright letters, suggesting that their behavioral changes resulted from recruiting mental rotation modules to perform the task rather than from modifications within the letter reading module. The human cognitive system employs different flexible computations to enable it to deal with changing demands in a changing world. Minimizing modifications and allocation of resources play a major consideration in the construction and implementation of the new computations.
To receive a reprint,
please send a request to:
dror@coglab.psy.soton.ac.uk