Abstract
Theories of visual information processing and feature integration, along with previous infant research, have indicated that only individual object features, and not the feature relations, are perceptually processed by early visual mechanisms. Previous research from within the memory domain has further suggested that, based on the encoding of features, errors in memory in the form of illusory conjunctions (i.e. miscombination of features from disparate objects) are likely. Considering that research has shown that infants' memory is particularly reliant on features, their recognition mechanisms should be quite susceptible to memory conjunction errors. Using the mobile-conjugate reinforcement paradigm, young infants were trained with mobiles displaying both O's and R's (or P's and Q's) and tested with stimuli that contained a miscombination of those features such as a Q (or an R), or a stimulus that did not contain these features (S). Results revealed that infants miscombined the individual features of the O and R (and P and Q) such that they recognized a Q (and R). Infants discriminated, however, the S mobile. These findings support previous research and the notion that object features encoded in infant long-term memory are likely stored as independent units irrespective of the spatial relations among those features, consequently allowing for the miscombination of features upon retrieval. Interestingly, however, infants discriminated a mobile displaying the same features as in the R but in a different spatial combination, suggesting that infants encoded some spatial characteristic of the individual feature. In sum, the present findings suggest that individual object features, and not the spatial feature relations, have a preferential status in young infants' long-term memory.