Using Multidimensional Encoding and Retrieval Contexts to Enhance Our Understanding of Stochastic Dependence in Source Memory

Abstract

Abstract In this chapter, we review and interpret research on multidimensional source memory paradigms. We focus in particular on research in which two source dimensions (e.g., color and location of a word stimulus), or features, or context attributes, are encoded and subsequently retrieved. We cover some of the basic empirical outcomes of interest, notably the finding of stochastic dependence among retrieval of source dimensions. Such dependence implies that retrieval of one source dimension is related to the retrievability of the other source dimension. Many research papers have established that stochastic dependence is supported by encoding factors that help bind items to context, and perhaps even context to context. Yet the finding of, and degree of, stochastic dependence at retrieval does not always relate predictably to other manipulations, such as when people are focused on one or the other source dimension in an encoding phase. Moreover, other theoretical proposals target retrieval processes, apart from encoding processes, as producing such dependence. Research that has included retrieval manipulations, such as external cuing of source dimensions at test, suggests that retrieval-specific causes of stochastic dependence have not been established. We also discuss other paradigms that are relevant to the retrieval of multiple object features and whether their retrieval is dependent or independent of one another. Mixed evidence results from this work, with some arguing for relative independence of object features, but others arguing for strong associative relationships among elements that predict somewhat all-or-none retrieval of items and their contexts or features (i.e., dependence). We attempt to integrate these various research paradigms and draw some broad, albeit tentative, conclusions about this collective work. We also suggest areas of research that should be considered as future work in multidimensional source memory.