Suppose that “memory task A” shows marked improvement at 5 months, but “memory task B” doesn’t show marked improvement until 9 months. Before we can make inferences about the development of memory, we need to understand how tasks A and B differentially strain the developing cognitive system.
Along these lines, Gross et al.’s 2002 Developmental Psychobiology article investigates the relationship of three different memory tasks in 6-month-old infants. The tasks are pretty representative of current behavioral work with human infants:
1) In the mobile conjugate reinforcement paradigm, infants have one of their legs tied to a mobile which moves every time they kick that leg. After a certain delay, infants are returned to the crib without having their leg tied to the mobile; rate of kicking above baseline is considered “memory” for the previous experience.
2) In the deferred imitation task, infants watch an adult manipulate an object, and some time later are given that object. The tendency for infants to imitate the previously observed actions (after a certain delay) is considered “memory”, relative to a control condition with infants who had not previously experienced the object.
3) In the visual recognition memory paradigm, infants are presented with a stimulus for a certain amount of time, and are (after a certain delay) later presented with that object and another novel object. The degree to which infants prefer the novel object, relative to control infants who had not seen either object previously, is considered memory for that object.
These tasks are theoretically interesting, insofar as the mobile task is often thought to index the development of an implicit, nondeclarative memory system, and the deferred imitation task is often thought to index explicit, declarative memory system. (The multiple-memory system framework is well-supported from evidence both with normal adults and those with acquired amnesia; see here for more on this). On the other hand, the visual recognition paradigm has been argued to index both systems. To the extent that these tasks show similar developmental trajectories, it becomes more likely that they measure neurobiologically-related processes.
Interestingly, Gross et al. found that 6-month-olds show evidence of retention on the mobile and deferred imitation task but not the visual recognition task. This result is the opposite of what might be \expected based on previous work! To confirm that this was not a spurious difference, the authors replicated the visual recognition task with different stimuli and with a procedure with better-matching contexts between training and test. In both cases, the same results were found.
Gross et al. conclude that the mobile conjugate and deferred imitation paradigms appear to measure the same underlying memory system, since age, retention interval, and changes in context all affect the two tasks. On the other hand, the authors suggest that the visual recognition task may index a different cognitive system.
Obviously, this conclusion needs to be tempered by a few considerations. First, even though the tasks were administered on the same day, it’s possible that they differed in motivational value. This could lead to spurious differences between tasks, even if they in fact rely on identical cognitive systems. (This problem could be attenuated if heart rate was measured during each task, as a proxy measure of arousal.) Secondly, there’s no “process-pure” task; any given memory task is likely to rely on multiple systems, even though those systems may reside in neurologically distinct areas. Therefore, a more cautious conclusion is that the mobile conjugate and deferred imitation tasks make similar demands on multiple memory systems, rather than relying on the same (single) system.
Finally, and most importantly, multiple tasks can be affected by the same factors and yet rely on different underlying systems. To make a strong claim that multiple tasks index the same cognitive system, each candidate task should be similarly affected by the interactions of factors like age, retention interval, and changes in context. This is the necessary step in linking a behavioral task to its underlying neural substrate, in the absence of other converging neural evidence (e.g., ERP, fMRI, or MEG imaging techniques). Despite these problems, the study clearly demonstrates that visual recognition memory is not comparable with deferred imitation and mobile conjugate reinforcement tasks, while the latter two tasks can demonstrate a remarkably similar profile at 6 months of age.