Reward-Motivated Learning and Memory Performance 107 

An important behavioral finding in both ERP-studies was that high rewards during learning promoted memory performance after a delay compared to low rewards, but only when pictures served as targets. That is, the behavioral reward effect was specific to the target-picture test blocks. This effect was primarily due to an increased correct rejection rate for non-targets (words) which had been studied with high reward. As already mentioned in Chapter III, this is thought to be in part because non-target words were easier to retrieve as they were identical in study and retrieval cue formats (copy cues) for this target designation (i.e., the non-target words were perceptually identical to their study representation). Additionally, memory is generally better for test materials that resemble the study material (Hornberger et al., 2004; Herron & Rugg, 2003). Thus, this finding might reflect less prioritization of target-pictures as a consequence of facilitated retrieval of non-target words because when both target-pictures and non-target words were exposed to high reward cues in this target-picture designation, encoding is likely to have provided a better memory

representation for these items. At test, therefore, this is manifest primarily by a more effective use of a recall-to-reject strategy for non-target items, in part due to the copy cue condition for this target designation. From this it might be inferred that motivational variables, like high monetary incentives, may provide better memory representations for both picture and word items, and that the retrieval of these representations is presumably further boosted for words because of the perceptual overlap from study to test. However, there was not a comparable boost in responding for these word items when they were designated as targets, which means that this assumption cannot account entirely for the findings of the work presented here. The target-picture specific reward effect might rather indicate that the extent to which copy cue presentation is beneficial depends upon the particular target designation.

On the one hand, a logical presumption would be to expect better overall memory performance after intentional than after incidental learning as participants during intentional learning were given the possibility to prepare themselves for effective encoding of the presented study items. However, this was not indicated by the current results presented in this thesis, as memory performances across the two conducted ERP-studies were highly similar. Comparison of the behavioral results of Experiment 1 and Experiment 2 in fact appeared to indicate a trend in the opposite direction (see Table 2.1 and Table 3.1). Whereas memory accuracy to targets was somewhat higher in Experiment 1 than in Experiment 2, accuracy to non-targets and new items was slightly lower. Statistically, however, there was no significant difference in accuracy and reaction times between the two conducted experiments (all p- values > .21). These similarities in memory accuracy across the two studies cannot be explained by the exclusion of more participants in Experiment 2 whose test performance was below chance level, as in both experiments the same number of participants were excluded (2).

On the other hand, the similarity in memory performance across both studies might be unsurprising if the following thought is taken into consideration. Even though participants in Experiment 1 were informed beforehand that their memory performance would be tested on the following day, they were not informed about either the specific aspects of the study items that would be most relevant to encode nor about the precise nature of the memory test (e.g. simple old/new recognition, recall or a memory exclusion test). This implies that even though participants in Experiment 1 may have used strategies to learn the items (e.g. by naming the pictures, grouping reward cue with study item, rehearsing), these strategies might not have been particularly appropriate for carrying out a memory exclusion task. Apparently and in accordance with Craik and Lockhart (1972), instructions or intentions to learn might be

efficient only to the extent that they generate suitable learning operations and consequently outperform those encoding processes that would be involved during incidental learning. Furthermore, there is even the possibility that individually chosen encoding strategies might hinder elaborative processing of new information needed for subsequent, successful memory retrieval. For example, in one of the earliest studies that compared memory performance after intentional and incidental learning by Eagle and Leiter (1964), participants were tested with a free recall test or a recognition test, either in an intentional or incidental learning condition. The authors observed that free recall test subsequent memory performance was facilitated in the intentional but not in the incidental condition. However, this effect was reversed after recognition testing. Memory performance here was instead superior after incidental learning. The authors’ main conclusion from these results was that optimal processing during encoding depends on how memory is tested afterwards. That is, whether intentional encoding outperforms incidental encoding with respect to later memory performance depends on factors such as the degree of effectiveness of the applied encoding strategy, characteristics of the study task and the type of retrieval task at test, because different memory tests require slightly different learning operations.

For participants that took part in Experiment 1 (and in Experiments 3a-d), the best way to earn or to keep the most money was to retrieve all items regardless of the level of incentives. But because of the relatively high task difficulty and the long retention interval between study and test (24 hours), a suitable compromise would have been to focus on highly rewarding or highly punishing items respectively. This means that the most efficient retrieval strategy would be to adapt retrieval processes specifically towards the high value items. By contrast in Experiment 2, a deliberately engaged strategic account on later memory performance cannot be made because firstly, the delivery of reward was related to performance in the study task and not to performance at test and secondly, the testing situation was incidental.