• Nem Talált Eredményt

4. Implicit versus explicit knowledge in the IL paradigms

4.2. The SRT task

Similarly to the controversy concerning the WP task, there is a long standing debate whether learning on the Serial Reaction Time task can be considered fully implicit. In one of the early SRT experiments, Willingham et al (1989) explored the role of conscious awareness on learning. In Experiment 1, participants faced four blocks of 100 trials. After the fourth block, participants were asked 1) whether they noticed something, and 2) whether they

noticed a repeating pattern throughout the experiment, and if so, then they were asked to replicate the pattern. Results showed that, out of the 60, 12 participants reported no knowledge of the sequence, 7 participants reported knowledge on the presence of the sequence, but were unable to provide any fragments of it. 29 participants reported the presence of the sequence, and were able to provide 4-9 element long fragments, while the remaining 12 participants were able to replicate the 10 element long sequence. Results

showed that overall RTs of the no-knowledge and some-knowledge groups did not differ from each other significantly, while RTs of participants with full explicit knowledge were

significantly lower than the other two groups on blocks 3 and 4 (with 100 trials in each

block). This experiment suggests that two thirds of the participants of an SRT study acquire at least partial explicit sequence specific knowledge.

Others point out that these values may even be higher. Willingham and colleagues (1989) used a free recall task for measuring conscious awareness: participants had to provide the sequence by their own. However, other results show that the free recall method is

insensitive (Shanks & St. John, 1994). Another method is cued recall, in which participants are shown a fragment of the sequence, and their task is to predict the next button press.

Results show that even participants who are categorized as unaware based on the free recall method may perform well above chance on the cued recall task (cf. Jackson & Jackson, 1995). This suggests that even more than two thirds of SRT task solvers may acquire some explicit sequence knowledge. A problem with the cued recall task though that in a number of cases, participants may respond above chance based on gut feeling (Destrebecqz &

Cleeremans, 2001). In such a case their response is contaminated by implicit knowledge (Neal

& Hesketh, 1997).

According to Destrebecqz and Cleeremans (2001), the major problem is the

assumption of process-purity: tasks are generally considered to rely purely on one system or

the other. The process dissociation procedure (Jacoby, 1991) takes into account both the automatic and intentional processes during memory retrieval. The PDP suggests the usage of two experimental paradigms in memory retrieval studies: inclusion and exclusion. The inclusion paradigm is the retrieval of the elements to be remembered, while the exclusion paradigm is generation with the avoidance of the target elements. Representations of the inclusion paradigm are facilitated by both automatic and intentional processes. In such a case, both automatic and intentional processes point to the same direction: that is, the activation of the given representation. On the other hand, in the case of the exclusion paradigm, the two processes interfere. Automatic processes facilitate, intentional processes inhibit the retrieval of the given representation. Application of the PDP to sequence learning (Destrebecqz &

Cleeremans, 2001) would suggest that both automatic and intentional processes work together in both free and cued-recall. On the other hand, if participants are required to provide a

sequence of button presses that do not overlap the previous repeating pattern, implicit knowledge may still facilitate the reproduction of the pattern, while intentional processes inhibit it. The greater the overlap between inclusion and exclusion paradigms, the less likely that explicit processes have much contribution to retrieval. Great overlap means that the activation of implicit knowledge is present, but participants have difficulties with the intentional inhibition of response structure.

In the SRT study of Destrebecqz and Cleeremans (2001), two conditions were

employed: a condition where there was no Response-Stimulus-Interval (RSI), and a condition where the RSI was 250 milliseconds. Results showed that both groups performed above chance on the inclusion test, while the groups only marginally differed from each other.

Earlier studies on this data would suggest that both groups show explicit sequence knowledge.

The exclusion paradigm showed a different picture though: participants of the RSI group were able to exclude sequence fragments during the generation task, whereas no-RSI participants

continued to generate chunks of the previously learned sequence, indicating no control over the expression of sequential knowledge.

In a recognition task, participants faced triplets that were either part of the previously learned sequence, or not (Destrebecqz & Cleeremans, 2001). Note that as second order predicative sequences were used, all target-pairs were part of the sequence, that is, only the third element defined the fragment as part of the repeating pattern or not. Results indicated that participants of the RSI condition could differentiate well between the two types of

fragments, whereas the confidence reports of the no-RSI participants were comparable for the two types. This confirms results of the exclusion criterion, that is, RSI participants have more conscious control over sequential knowledge, while mostly implicit processes underlie learning in the no-RSI condition.

Results of previous studies concerning the explicit versus implicit nature of the SRT task are mixed. It seems to be clear however that participants are very likely to acquire some kind of explicit sequence knowledge. This however may differ by procedure (Destrebecqz &

Cleeremans, 2001), as well as motivation and task-difficulty (Fu, Fu, & Dienes, 2008). The most important aspect of the above introduced studies is that the SRT task cannot be

discussed in the framework of process purity. Similarly to the WP task, the SRT task cannot be considered a purely implicit task. This assumption is in concert with the reviewed

neuropsychological and imaging results.