MODALITY INDEPENDENCE OF IMPLICITLY LEARNED GRAMMATICAL KNOWLEDGE

In each of 4 experiments exemplars of an artificial grammar were presented in one modality (e.g., sequences of tones differing in pitch in Experiments 1 and 2, or sequences of spoken syllables in Experiment 3, or sequences of arbitrary graphical symbols in Experiment 4), but the subsequent classification task was performed on novel sequences in another (e.g., sequences of letters in Experiments 1 and 2, sequences of arbitrary graphic symbols in Experiment 3, and sequences of written syllables in Experiment 4). Prior exposure to the grammar improved classification of novel stimuli, even across modalities. Participants who received either no preexposure or were exposed to pseudorandom sequences showed no such improvement. Consequently, part of the learning process can take place prior to any exposure to the domain within which categorization is to take place.