New implicit memory models

The observation that performance can be increased by a prior experience without explicit knowledge of that experience, led to the development of theories concerning what was termed implicit memory. Much research has since been completed and many theories have been posited to explain implicit memory. The issues surrounding the nature of implicit memory are explored here in four studies using models that aim to advance the understanding of the processes involved. The first two studies explore implicit memory as measured by the process dissociation task (Jacoby, 1991). These studies find that the process dissociation task lacks the information to properly measure implicit and explicit memory. Further the models traditionally used with this method are found to be fatally flawed, because of problematic assumptions and inadequate information. As a result of these problems the process dissociation models and task are rejected in favor of a new implicit memory model. This storage-based implicit memory model is based on model 7B (Chechile, 2004) and the four-alternative task used by Chechile and Soraci (1999). The use of this model in two studies, Experiments 3 and 4, highlights the interrelated nature of implicit and explicit memory. Experiment 3 explores the differences in these processes by manipulating the level of encoding via a read-generate manipulation, finding both implicit and explicit memory increasing with more effortful encoding. Experiment 4 examines the effect of retention interval for these memory processes. Experiment 4 reveals decreases in explicit memory as lag increases and increases in implicit memory within this interval, suggesting that explicit and implicit memory are not separate systems. These findings support the interpretation of implicit memory as a different level of storage distinct from other forms of storage. Taken together these studies suggest that implicit storage may be differentiated from explicit and fractional storage. That is implicit memory is likely a residual memory representation rather than a separate memory system. Further these findings are consistent with prior research using multinomial processing trees to model storage, knowledge and guessing processes.