Running head: IMPLICIT LEARNING

Ruth C. Capistrano Implicit Learning University of North Florida

IMPLICIT LEARNING Abstract Forty-nine undergraduates participated in a study of implicit learning. Consistent with previous research, implicit learning occurred as demonstrated by a significant decrease in reaction time from Block 1 to Block 2 and a corresponding increase in reaction time when the underlying structural relationship of the sequential stimuli was changed. The results suggest that shorter reaction times demonstrate the subjects ability to unconsciously predict stimuli locations; in other words, they implicitly learned the stimulis sequential pattern of appearance. Keywords: implicit learning, unconscious processes

IMPLICIT LEARNING Introduction Environmental knowledge is instrumental in recognizing broad and specific regularities of ones environment to orient oneself. But most importantly, this knowledge, once manipulated, determines how one reacts to his or her immediate surroundings. The acquirement of such knowledge involves explicit and implicit processes. As nominally implied, explicit learning yields consciously derived information whereas implicit learning detects environmental regularities unconsciously (Yordanova, Verleger, Wagner, & Kolev, 2010). Implicit learning is particularly helpful in decision-making within novel environments and the limitations of neuroclinical disorders (Reber, 1989). A common measure of implicit learning is artificial grammar learning (AGL). Subjects are presented with a string of stimuli (numbers, letters or symbols) arranged in a statistically determined pattern that is not overtly recognizable. The subjects first undergo training, in an attempt to acquire the knowledge necessary to master the grammar, but its abstract structure can only be learned implicitly, because it is far too complex to learn in one experimental sitting. They are then subjected to an assessment of some form, to test what they did in fact learn. Researchers use various assessments depending on what they are interested in revealing; a

common form, however, is the well-formedness task. In this task, the subjects are presented with a stimulus string and respond whether or not the string falls under the grammar by following the its definitive rules. As mentioned earlier, knowledge that is processed implicitly is structured i. e., a complex rule dictates the stimuli pattern, and abstract, meaning that the stimuli sequence is dependent upon the underlying relationship (rule), and not its physical appearance. To demonstrate the

IMPLICIT LEARNING perquisite importance of structure, Reber presented rule-governed stimuli to one group of subjects and non-ordered stimuli to the other (1967). Both groups believed the experiment dealt with rote memory and not implicit learning so they did not experience grammar training but merely observed the exemplars (stimuli strings). The structured stimuli group memorized the

exemplars significantly better than the non-structured stimuli group. Furthermore, the successful group correctly distinguished novel stimuli strings that followed the rule from those that did not. In other words, implicit learning only occurred with structured stimuli and failed to manifest with random, non-structured stimuli. Abstraction, the other prerequisite component to implicit learning, is exhibited in an AGL study conducted by Reber (1969). In the training phase, the subjects memorized strings of letters from a synthetic grammar. In the testing phase, however, only one experimental group made use of the training exemplars. In their case, the underlying relationship of the grammar did not change but the aesthetics diddifferent letter stimuli replaced the letters from training. For the other group, training became meaningless since a new relationship altogether created the letter strings. In this instance, the letters were the same; the testing trials were physically indistinguishable from the training trials but in actuality, the two trial sets fundamentally had nothing in common. The two control groups included the condition where nothing changed (identical training and testing trials) and the condition where both aspects, the rules and aesthetics, altered. Reber found that the change in physical appearance of the letter strings had no bearing on performance, but an overhaul in the relational rules of the stimuli led to a lowered level of performance. This study reveals that implicit learning cannot be explained by the appearance of a stimulus, but instead must be accounted for by its abstract aspect, the underlying relationship. Abstraction allows one to apply certain rules in novel environments. In AGL

IMPLICIT LEARNING

experiments, for example, abstraction prevents a subject from relying purely on the regurgitation of exemplars. The successful approach to novelty is supported in a compelling theory that states the origin of implicit learning is grounded in the survival of the individual and, ultimately, species. Several studies look at disorders where explicit processes expired but implicit processes are still intact. In Warrington and Weiskrantzs study, subjects with amnesia did not perform well in overt word recall and recognition assessments but carried out the implicit knowledge-recruiting assessments of word-stem and word-fragment cues sufficiently (1982). The advantage of implicit learning over explicit learning is seen in another study where depressives, schizophrenics, and alcoholics with brain damage all yielded significantly better results in the implicit AGL task than in the explicit short-term memory task (Abrams & Reber 1989). Destrebecqz & Cleeremans (2001) measured implicit learning in a different fashion than the AGL. Instead, they used a task that is the basis for the experiment of the present study. In each trial, a stimulus is presented in one of four possible locations. The sequential locations seen from trial to trial within a block are patterned according to a rule-governed order termed the Second Order Conditional (SOC). The subjects are exposed to two sequential patterns i.e., two variations of the SOC, and divided into two groups. Both halves received one variation in Blocks 1-12, 14, and 15 and received the other variation in Block 13. In the inclusion instruction set version of the experiment, the subjects pressed a key at the appearance of the stimuli, as fast as they could. The reaction times are then recorded. The results indicated slower (or longer) reaction times in Block 13, where the subjects received the variation alternative to the variation they were trained in from Blocks 1-12. In the following experiments of this study, the subjects attempted and failed to reproduce the stimuli pattern, retook the first experiment but instead

IMPLICIT LEARNING avoided the stimuli locations; and completed a stimuli recognition test, in which they subjectively rated the familiarity of the covert patterns. The purpose of the present study is to see if the results will echo that of Destrebecqz & Cleereman (2001), garnering additional evidence for the phenomenon of implicit learning. Method Participants The participants were 13 undergraduates (3 males) enrolled in a Human Learning and Performance course at the University of North Florida. Participation satisfied a partial course requirement. We combined their data with data from four prior semesters, resulting in 49 participants. Apparatus Each participant worked at a Dell Optiplex 780 PC running under Windows 7, equipped with a standard keyboard, monitor, and mouse. CogLab 2.0 on a CD (Francis, Neath, and VanHorn, 2008) presented the stimuli, and recorded the responses and response times. Procedure Each participant sat approximately 2 ft. from his or her monitor, with fingers resting on the z, x, ., and / keys. The participant pressed the space bar to initiate each trial.

On each trial, a red-outlined oval appeared in one of four possible locations: the left-most outer location, inner left location, inner right location, and right-most outer location. The task was to press the spatially appropriate key as fast as possible. Thus, if the oval appeared in the

IMPLICIT LEARNING left-most location, the correct response was to press the left-most key, the z-key; if the oval

appeared in the inner left location, the correct response was to press the inner left key, the x-key; similarly, an oval appearance in the inner right location corresponds with the .-key and an oval appearance in the right-most outer location corresponds with the /-key, the right-most outer key. The stimulus remained onscreen until the participants response. CogLab presented 288 trials with an option for a brief break every 24 trials. Each participant completed 11 blocks of trials. Design The present study followed a single-factor repeated measures design. The independent variable was Block (1-11). The dependent variable was the time (in msec) between the onset of the stimulus and the participants correct response. Results Table 1 presents the Analysis of Variance for Blocks. As the table shows, we found a main effect of Blocks (F(10, 460) = 11.28, p < .05). Table 1. Analysis of Variance for Blocks. Source Subjects Blocks Error Total SS 3039941 256793 1047496 4344230 df MS 46 66085.68 10 25679.28 460 2277.17 516 F 11.28

Table 2 presents the means and standard errors for Block. As the table shows, Block 6 yielded the fastest response (450 msec) and Block 1 yielded the slowest response (531 msec).

IMPLICIT LEARNING Also, RT decreased the most from Block 1 (531 msec) to Block 2 (473 msec). There was an

increase from Block 6 (450 msec) to Block 7 (465 msec), immediately followed by a decrease to Block 8 (451 msec). Table 2. Mean RT (msec) and standard errors for Block. Block 1 2 3 4 5 6 7 8 9 10 11 Mean 531 473 456 455 454 450 465 451 465 478 485 SE 16 13 11 12 11 11 15 12 13 12 17

Figure 1 presents the same data in graphic form.

540 520 Mean 500 RT 480 (msec) 460 440 420 1 2 3 4 5 6 7 Block 8 9 10 11

Figure 1. Mean RT (msec) as a function of Block.

IMPLICIT LEARNING The planned comparison between Block 1 and Block 2 shows a significant difference t (490) = 5.85, p < .05. The planned comparison between Block 9 and Block 10 also shows a significant difference t (490) = 1.27, p < .05. Conclusion The subjects completed 12 blocks but the researchers discarded the results of the first block to avoid the confounding variable associated with adjusting to the experiment. The main

effect found for Blocks indicates that implicit learning occurred over the span of 11 blocks. The differences in mean reaction time (RT) reveal that the learning did not take place at a steady, continuous rate. Instead, there are some blocks where implicit processes are more engaged than in others. The drastic decrease in RT from Block 1 to Block 2, for example, demonstrates the occurrence of implicit learning whereas the increase in RT from Block 6 to Block 7 reveals the inefficacy of implicit learning. The disparity in RT is so different that this difference can only be accounted for by implicit learning. The significant planned comparison validates this statistical difference. The quicker or shorter RT exemplifies implicit learning because the subjects could unconsciously predict where the upcoming stimulus location would be; in other words, they implicitly learned the stimulis sequential pattern of appearance. The researchers expected an increase in RT from Block 9 to Block 10 because the CogLab program was anticipated to change the sequential rules and underlying relationship. A decrease in RT from Block 10 to Block 11 was supposed to have followed, since the Block 11 would be a return to the original sequential rules. It appears that the software program altered and returned to the original relationship in Block 6 and Block 7, respectively. Thus, the technical difficulties of the software prevented the expected results from manifesting. In

IMPLICIT LEARNING

10

addition to the technical difficulties a flaw in the experiment is the failure of CogLab to state the rules the sequential stimuli followed. To conclude, implicit learning took place because the experimental stimuli had an underlying abstract and structural relationshipthe sequence followed a rule of some kind that could not be overtly recognized and decoded. An improvement upon the experiment would be a post test, like the postrecognition test given in Destrebecqz and Cleeremans study (2001), that can empirically validate the efficacy and leverage implicit processes have over explicit processes.

IMPLICIT LEARNING References Abrams & Reber, A. S. (1989). Implicit learning in special populations. Journal of Psycholinguistic Research, 17, 425-439.

11

Destrebecqz, A. & Cleeremans, A. (2001). Can sequence learning be implicit? New evidence with the process dissociation procedure. Psychonomic Bulletin & Review, 8, 343-350. Francis, G., Neath, I., & VanHorn, D. (2008). CogLab 2.0 on a CD. Belmont, CA: Wadsworth, Cengage Learning. Reber, A. S. (1967). Implicit learning of artificial grammars. Journal of Verbal Learning and Verbal Behavior, 77, 317-327. Reber, A. S. (1969). Transfer of syntactic structure in synthetic languages. Journal of Experimental Psychology, 81, 115-119. Reber, A.S. (1989). Implicit learning and tacit knowledge. Journal of Experimental Psychology: General, 118, 219-235. Warrington, E. K., & Weiskrantz, L. (1982). Amnesia: A disconnection syndrome? Neuropsychologia, 20, 233-248. Yordanova, J., Verleger, R., Wagner, U., & Kolev, V. (2010). Patterns of implicit learning below the level of conscious knowledge. Journal of Psychophysiology, 24, 91-101.