ABSTRACT
When participants are asked to learn letter strings, which were constructed on the basis of a complex rule system (an artificial grammar), they are able to classify novel letter strings as being grammatical or nongrammatical better than chance without explicit knowledge about the rules. We tested whether violations of such complex regularities can be detected by the brain, when strings were presented sequentially (i.e. letter by letter). Compared with regular letters, rule-violating letters elicited enlarged amplitudes of the N1 component in the event-related potential, indicating that violations are automatically detected by the brain. However, this effect occurred irrespective of the participants' classification of the strings, indicating that the brain's detection of regularity violations does not necessarily lead to correct classifications.
