Implicit and explicit learning of event sequences: evidence for distinct coding of perceptual and motor representations

Acta Psychologica 104 (2000) 45±67

www.elsevier.com/locate/actpsy

Implicit and explicit learning of event sequences: evidence for distinct coding of perceptual and motor representations Jascha R usseler *, Frank R osler Experimental and Biological Psychology, Department of Psychology, Philipps University, Gutenbergstr 18, D-35032 Marburg, Germany Received 12 June 1999; received in revised form 27 August 1999; accepted 2 September 1999

Abstract Event-related brain potentials (ERPs) of 21 subjects were recorded in a choice reaction time task with a repeating eight-element long stimulus sequence. The regular event sequence was sometimes interrupted by `perceptual' or by `motor deviants' which both replaced an expected stimulus but either preserved or violated the sequence of motor responses. Response times con®rmed that all subjects had acquired some knowledge of the sequential dependencies. By means of a post-experimental free recall and recognition test, subjects were classi®ed as having either explicit or implicit knowledge of the event sequence. The ERPs showed di€erent e€ects for di€erent types of stimuli and the two groups. In the group of explicit learners, a larger N200 component was evoked by both types of deviants and a larger P300 by motor deviants only. In the group of implicit learners these `perceptual components' remained una€ected. In contrast, in both groups of subjects the lateralized readiness potential (LRP) which accompanied motor deviants revealed a partial activation of the to be expected but incorrect response, i.e. motor learning. These results suggest that explicit learners acquire knowledge about both, stimulus and response dependencies while implicit learners acquire knowledge about response dependencies only. Ó 2000 Elsevier Science B.V. All rights reserved. PsycINFO classi®cation: 3297; 2560; 2343

*

Corresponding author. Present address: Clinical Neuropsychology, Department of Psychology, Otto-von-Guericke University, Lennestr. 6, D-39112 Magdeburg, Germany. Tel.: +49-391-67-14845; fax: +49-391-67-14815. E-mail addresses: [email protected] (J. RuÈsseler), [email protected] (F. RoÈsler). 0001-6918/00/$ - see front matter Ó 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 1 - 6 9 1 8 ( 9 9 ) 0 0 0 5 3 - 0

46

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

Keywords: Event-related potentials; N200; P300; LRP; Sequence learning

1. Introduction The goal of this study is to provide evidence that implicit and explicit learning of event sequences involve di€erent processing systems. To objectify this claim, we recorded event-related brain potentials (ERPs) while subjects performed a variant of the serial reaction time (SRT) task. In this task, subjects are confronted with a sequence of stimuli (x's or *) appearing at di€erent locations on a computer screen and are instructed to press a corresponding key for each location as fast and as accurate as possible. Unknown to subjects, the stimuli appear in a regular repeating sequence of positions (e.g. 4-2-3-1-3-2-4-3-2-1, with 1 corresponding to the leftmost and 4 to the rightmost position of a horizontally aligned display; Nissen & Bullemer, 1987). After several structured `training' blocks subjects are transferred to an unstructured stimulus sequence. Typically, subjects show a prolongation of reaction time (RT) in the unstructured compared to the preceding structured block which is taken as evidence that the sequential stimulus structure was learned. Sequence learning was found without the concurrent development of conscious awareness for the sequential structure of the stimulus material (e.g. Cherry & Stadler, 1995; Cohen, Ivry & Keele, 1990; Curran & Keele, 1993; Eimer, Goschke, Schlaghecken & St urmer, 1996; Frensch & Miner, 1994; Howard & Howard, 1989, 1992; Mayr, 1996; Nissen & Bullemer, 1987; Reed & Johnson, 1994; Stadler, 1992, 1993, 1995; Willingham, Nissen & Bullemer, 1989). However, it is still an open question which types of representations are formed during implicit learning (for reviews, see Clegg, DiGirolamo & Keele, 1998; Curran, 1998; Goschke, 1998; Ho€mann & Koch, 1998). The available evidence is contradictory, indicating either learning of response±response (R±R), stimulus±stimulus (S±S) or stimulus±response (S±R) associations. Willingham et al. (1989) suggested that associations between stimuli and responses (S±R learning) are of primary importance for the acquisition of sequence knowledge. In their study, subjects responded to the color of stimuli appearing at di€erent locations. Subjects failed to show an RT-advantage for structured blocks, if the task-relevant sequence of colors and responses was unpredictable although the stimulus-locations followed a predictable sequence. In contrast, if the sequence of colors and the related responses were predictable but the stimuli appeared at randomly determined locations performance improved. However, if subjects were instructed to respond to the location of uncolored stimuli which followed the same sequence as before no transfer was found. Thus, the authors concluded that stimulus structures are learned only if they are relevant for subsequent behavior and if they can be mapped directly onto motor responses. Howard, Mutter and Howard (1992) found that subjects who simply observed sequentially structured stimuli learned as much as subjects who responded to the stimuli with key-presses throughout the learning phase. In a more recent study, Mayr

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

47

(1996) found that spatial sequences could be learned independently of response sequences (see also Cohen et al., 1990; Keele, Jennings, Jones, Caulton & Cohen, 1995; Stadler, 1989). These ®ndings indicate learning of stimulus±stimulus associations. Finally, Nattkemper and Prinz (1997) obtained evidence in favor of a motor learning perspective (R±R learning). In their experiments, pairs of stimuli were always assigned to one response. Unexpected manipulations of the stimulus sequence which did not interrupt the response sequence were not accompanied by an RT increase whereas violations of both, stimulus and response sequences, resulted in a prolonged response latency. Converging evidence that implicit learning in the SRT task is a type of motor learning was also provided by recent positron emission tomography (PET) studies. These studies suggest that implicit and explicit learning involve di€erent neural systems. During implicit sequence acquisition an increase in regional cerebral blood ¯ow (rCBF) was found in motor areas, i.e. in the sensorimotor cortex, the supplementary motor cortex and in the basal ganglia (Doyon, Owen, Petrides, Sziklas & Evans, 1996; Grafton, Hazeltine & Ivry, 1995; Hazeltine, Grafton & Ivry, 1997). In contrast, during explicit learning, enhanced activity was found in non-motor areas as the right dorsolateral prefrontal cortex, the right premotor cortex, the right ventral putamen, and the biparietal±occipital cortex (Grafton et al., 1995). In the present study, we used ERPs to explore whether functionally di€erent processes contribute to explicit and implicit learning of event sequences. In particular, we want to provide converging evidence to the PET ®ndings that motor processes are of primary importance for implicit sequence learning whereas motor as well as perceptual processes are relevant for explicit sequence acquisition. ERPs seem to be especially suited to substantiate this perspective, because di€erent ERP components are known to re¯ect either perceptual and stimulus-evaluation processes or response preparation processes, respectively. Moreover, ERPs re¯ect a completely di€erent type of signal than PET. ERPs are evoked by electrical rather than blood ¯ow changes and they are coupled much more directly to the processing of single events, because they can be measured during the short epoch which extends between stimulus presentation and response execution. 1.1. ERPs of perceptual and stimulus evaluation processes Task relevant stimuli of low probability in an otherwise regular sequence of events elicit an enhanced negativity with onset of about 200 ms post-stimulus (N200 component). This omponent is followed by an enhanced positivity with onset latency of about 350 ms (P300 component; e.g. Courchesne, Courchesne & Hillyard, 1978; Duncan-Johnson & Donchin, 1982; Gehring, Gratton, Coles & Donchin, 1992; Ritter, Simson & Vaughan, 1983; Squires, Donchin, Herning & Mc Carthy, 1977). Both components seem to re¯ect stimulus-evaluation processes. The amplitude of the N200 component was found to be inversely related to the probability of either attended or unattended infrequent stimulus changes (for a review, see Ritter et al., 1984). P300 amplitude was found to be sensitive to both the subjective stimulus probability and to the task relevance of the presented material (e.g. Gehring et al.,

48

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

1992; Matt, Leuthold & Sommer, 1992; Sommer, Matt & Leuthold, 1990; Squires, Donchin & Herning, 1977; for reviews see Donchin & Coles, 1988; Johnson, 1988). 1.2. ERPs re¯ecting response preparation The lateralized readiness potential (LRP) is an indicator of response selection and response activation (Coles, 1989). It is computed from the readiness potential (RP), a slow negative-going potential that starts some time before a movement and rises gradually to its maximum just before movement onset (Kornhuber & Deecke, 1965). The RPs preceding voluntary ®nger and hand movements are larger contralateral to the performing hand (Kutas & Donchin, 1980). To exclude asymmetries which are unrelated to the motor response De Jong, Wierda, Mulder and Mulder (1988) and Gratton, Coles, Sirevaag, Eriksen and Donchin (1988) suggested to average the RP asymmetries obtained for left- and right-hand movements. To this end, the activity of contra- and ipsilateral electrodes is ®rst subtracted point by point. The two resulting di€erence waves are averaged to obtain the LRP, which re¯ects the net asymmetry of the RP preceding lateralized hand or ®nger movements. Several ®ndings qualify the LRP as a speci®c index of response preparation. First, part of the LRP seems to be generated in the precentral motor cortex contralateral to the activated muscle group (see Sommer, Leuthold & Ulrich 1994). Second, numerous studies demonstrated a systematic relationship between the LRP and response selection (e.g. Gratton et al., 1990; Gehring et al., 1992; Hackley & Miller, 1995; Miller & Hackley, 1992; Osman, Bashore, Coles, Donchin & Meyer, 1992; Osman, Moore & Ulrich, 1995; Osman & Moore, 1993). For example, Gratton et al. (1988) asked subjects to respond as fast as possible to one of the two imperative stimuli primed by a warning tone either with their left or their right hand. They found a relationship between correctness of the responding hand and the polarity of the prestimulus LRP for fast responses. Correct responses were preceded by a negative-going LRP while incorrect responses were preceded by a transient positive LRP amplitude (positive `dip'). 1.3. The present study In the present study, we used these ERP measures to explore the processes underlying sequence learning and to test the hypothesis that explicit and implicit learning di€er in the neural systems involved. To this end, we modi®ed the SRT paradigm as follows: A set of eight di€erent letters was used as stimulus material (see also Eimer et al., 1996). Subjects had to respond to a particular letter by lifting one of the four ®ngers. Two di€erent letters were always related to one response ®nger (Nattkemper & Prinz, 1997). This arrangement allowed us to introduce two types of deviant stimuli in the otherwise regular sequence: Perceptual deviants were created by switching between the two letters which required the same response. Thus, the sequence of perceptual events was changed but the sequence of responses was preserved. Motor deviants were created by exchanging a regular letter with one which

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

49

required a response with the opposite hand. Motor deviants were, of course, also perceptual deviants. If the system encodes and stores the perceived stimulus sequence, then any deviation from the regular stimulus sequence should become apparent as an enlarged N200 or P300 which re¯ects stimulus evaluation processes. If the system also encodes the sequence of motor acts, then the LRP should be a€ected as well, that is, an initial negative-going LRP in standard and perceptual deviants and an initially positivegoing LRP for motor deviants (positive-going `dip') are expected. With respect to the three hypotheses ± S±S learning, S±R learning and R±R learning ± the following outcomes can be predicted: For pure S±S learning, we expect an amplitude change in the N200 and P300 components but no activation of incorrect responses for motor deviants in the LRP (positive-going `dip'). For pure R±R learning, we expect the opposite outcome: No amplitude changes for N200 and P300, but an activation of the incorrect but expected response-hand for motor deviants. Finally, if both processes contribute to sequence learning (S±S as well as R±R associations) we expect both amplitude changes of the N200 and P300 and an LRP `dip' for motor deviants. Recent neuroimaging research with PET suggests that implicit and explicit learners process sequence knowledge di€erently (e.g. Grafton et al., 1995). In particular, the motor system seems to be of prime importance for implicit learning while perceptual systems seem to contribute to explicit learning. These processing di€erences should become manifest in the ERP e€ects, too. We expect that implicit learners, who may acquire `motor knowledge' only, show an LRP e€ect but no amplitude changes of the perceptual components N200 and P300. In contrast, explicit learners, who may also acquire knowledge of the stimulus sequence should show N200 and P300 amplitude changes and an LRP dip. If learning of S±S associations is an important component of sequence learning, then RTs for perceptual as well as for motor deviants should increase relatively to RT to standard letters. In contrast, if only R±R associations are learned, we expect an increased RT for motor but not for perceptual deviants. If both processes contribute to learning, we expect an increase of RT for perceptual and an additional RT prolongation for motor deviants. 2. Method 2.1. Subjects In total, 21 subjects participated. Data of two subjects had to be discarded because of extensive ocular artifacts. The ®nal sample comprised 12 female and 7 male subjects between 20 and 36 years of age. According to self-report, two subjects were left-handed. All participants had normal or corrected-to-normal vision. All but two were students of the Philipps-University, Marburg. Participants either received course credit or were paid on an hourly basis. No subject had participated in other sequence-learning experiments before.

50

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

2.2. Stimuli and apparatus Subjects were seated in an electrically shielded, sound attenuated and dimly lit room. Eight capital letters (K, L, M, R, S, T, V and X) served as stimuli. The letters were presented at the center of a computer display (Atari SM 124, refresh rate 72 Hz) located in front of the subject. The letters appeared in black on a white square subtending a visual angle of 1.45°. Letters subtended visual angles of 0.58° (height) and 0.28° (width) at a constant viewing distance of 100 cm. Each letter remained on the screen until a response had been given by the subject. Subjects placed their left and right middle and index ®ngers in four circular cavities each equipped with a light gate. To respond to a stimulus, the respective ®nger had to be brie¯y lifted from the cavity. RTs were measured from stimulus-onset to the subjects' response to the nearest 5 ms. The response-to-stimulus interval was held constant at 500 ms. This was done to make the task as comparable as possible to other sequence-learning studies (e.g. Nissen & Bullemer, 1987). If there was no response within 5 s the stimulus disappeared and the trial was counted as an error. The letters were related to the response keys in the following way (see Fig. 1): For M and T, subjects had to respond with their left middle, for V and R with their left index, for X and K with their right index and for L and S with their right middle ®nger, respectively. The experiment comprised 38 blocks of 96 letters each. In blocks 1±4, 20 and 36 the letters were presented in a pseudorandom, unpredictable sequence. All other blocks were regular, i.e. the stimulus sequence was predictable. Regular blocks were constructed from an eight-element letter sequence (V L K T X S M R). According to Cohen et al. (1990), this sequence is of unique statistical structure. The corresponding ®nger lift sequence is I m i M i m M I (M and I denote middle and index ®ngers of the left hand, m and i middle and index ®ngers of

Fig. 1. Mapping of letters to responses. Arrows indicate examples of deviant stimuli.

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

51

the right hand). Thus, the response sequence is more complex than the stimulus sequence in the sense that two preceding responses must be remembered to correctly anticipate the next response alternative (hierarchical structure according to Cohen et al., 1990) whereas only one letter is needed to correctly predict the next stimulus. Each regular block was initiated with a randomly selected letter and the following letters were then fully determined by the sequence (e.g.TXSMRVLKTXSMR. . .). However, within each of the 12 sequence repetitions of one block one regular letter was always replaced by one of the two di€erent types of deviant letters: Perceptual deviants changed the perceptual event sequence but left the response sequence unchanged. To achieve this a regular letter was replaced by the second letter related to the same response (e.g. the letter M was replaced by T, both requiring a response with the left middle ®nger). Motor deviants changed both the perceptual and the response sequence. In this case, a regular letter was replaced by any of the four letters which required a response of the opposite hand (e.g. the letter M was replaced by either X, K, L or S; see Fig. 1). The position of the deviant letter was determined randomly. In pseudorandom blocks, stimuli were determined randomly with the exception that in a series of eight subsequent stimuli, each letter had to occur once. Thus, the probability of each letter was the same in regular and pseudorandom blocks. For the recognition procedure (see below), bigrams, trigrams and quadrupels of letters were constructed. One half of these test strings was identical to original sequence fragments, the other half was identical in all but one element. We used 10 bigrams, trigrams and quadrupels each. Five of the fragments of a given length were not part of the original eight-letter sequence (see Appendix A). 2.3. Procedure After electrode montage, subjects started with practice blocks to learn the relation between stimuli and responses, until they completed one block of 96 trials with less than six errors. Subjects of the two post-experimentally formed groups of explicit (E) and implicit (I) learners (see below) did not di€er with respect to the number of performed training blocks (E: 4 blocks, I: 3.8 blocks on average, respectively, t ˆ 1:53, P > 0.145). Letters were presented randomly in these blocks. After each block, subjects received feedback about the number of erroneous responses and mean RT. Accuracy and speed were both stressed in the instruction. Subjects started the next block by placing their ®ngers in the circular cavities. Upon completion of the 38 experimental blocks, subjects had to answer a postexperimental questionnaire comprising four questions and three rating scales. In question 1, subjects had to state whether they had noted any structural regularity in the stimulus material (yes/no-response). Question 2 asked if the letters had appeared in a random or in a predictable order. Subjects were now informed about the presence of a stimulus sequence and were asked to report everything they had noticed about the regularities of the letter sequence. In question four participants were asked to freely reproduce the letter sequence (free recall procedure). Questions 5±7 formed the recognition test. Subjects were given 10 bigrams (question 5), 10

52

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

trigrams (question 6) and 10 quadrupels (question 7) of letters. For each of these letter sequences they had to indicate on a ®ve-point rating scale whether it had been part of the stimulus sequence in the foregoing experiment. After completing the questionnaire, subjects were fully debriefed about the purpose of the study and the electrodes were removed. The experiment extended over 3±4 h. 2.4. EEG recording, artifact rejection and signal extraction The EEG was recorded from 61 Ag±AgCl electrodes placed on the subjects' head by means of an elastic cap (Gaggl-system, Graz, Austria). Electrodes are labeled according to a modi®ed version of the 10±20 system of electrode placement (Jasper, 1958). The cap was positioned on the head with reference to the nasion, inion and the preauricular notches so that the vertex electrode (Cz) was positioned correctly. Prior to electrode ®xation individual scalp-sites were cleaned and abraded through holes in the cap designed to ®xate the electrodes. Electrodes were ®xated on the cap after injection of a conduction gel (Synapse by Med Tek). All scalp-electrodes were referenced to linked earlobes. To control for vertical and horizontal eye movements, the electrooculogram (EOG) was recorded from the outer ocular canthi (horizontal EOG) and the sub- and supraorbital ridges (vertical EOG), respectively. Impedances of all electrodes were kept below 5 kX. Two sets of 32-channel ampli®ers (SYNAMPS) were used for EOG and EEG recording with a digitization rate of 100 Hz. Bandpass ®lters were set from DC to 50 Hz. An ATARI Mega ST2-computer controlled stimulus presentation as well as behavioral and electrophysiological data collection. The EEG signals were stored by a 486 PC running DOS and NEUROSCAN acquire software. Prior to the beginning of each experimental block a DC-reset was initiated automatically. EEG was averaged o€-line for epochs of 1000 ms, starting with the presentation of the stimulus and ending 1000 ms after stimulus-onset. An individual pre-stimulus baseline ()200±0 ms) was used. Epochs were averaged separately for each experimental half (1st half: blocks 5±19, 2nd half: blocks 21±35, 37,38) and stimulus type (standards, perceptual deviants, motor deviants; the pseudorandom blocks were excluded). Prior to averaging, trials with overt response errors, ocular or muscular artifacts were rigorously rejected (rejection criterium: amplitude >50 lV at electrodes vEOG, hEOG or Cz). We also computed separate LRPs for each stimulus type, experimental half and subject. For computation of stimulus-locked LRPs we used epochs beginning 500 ms prior to and ending 1000 ms after stimulus-onset relative to a baseline 500±400 ms prior to stimulus presentation. This early baseline was chosen as we wanted to see whether an anticipatory response selection will take place. We ®rst computed separate averages for correct left- and right-hand movements. Second, the di€erence of contra- and ipsilateral electrodes was calculated (for left-hand movements: C4±C3, for right-hand movements: C3±C4). Finally, these two waveforms were averaged. To summarize, the LRP was computed according to the following

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

53

formula: LRP ˆ [Mean(C4±C3)left-hand movement + Mean(C3±C4)right-hand movement ]/2 (see Coles, 1989). Negative de¯ections in the resulting LRP waveform indicate activation of the correct hand whereas positive de¯ections indicate activation of the incorrect hand. 2.5. Dependent variables and statistical analysis Groups were formed on the basis of post-experimental questionnaire results. We computed the percentage of correctly generated elements in the free-recall procedure and a recognition score for bigram, trigram and quadrupel ratings. Answers on the ®ve-point rating scale were scored as follows: For items which had actually been presented in the sequence 2 and 1 points, respectively, were assigned to the ratings `totally con®dent' and `fairly con®dent that the letters had been part of the sequence', 0 points for the rating `do not know', and )1 and )2 points, respectively, to the ratings `fairly con®dent' and `totally con®dent that the letters had not been part of the sequence'. For items which had not been part of the stimulus sequence scoring was reversed. This results in a score varying between )20 and 20 points for bigrams, trigrams and quadrupels, respectively. Increasing positive values indicate increasingly explicit knowledge of the sequence as assessed by a recognition test. Zero or negative values indicate no explicit knowledge. The three scores were added to one overall `recognition score'. Error rates and mean RTs were determined separately for standard, pseudorandom, perceptual and motor deviant letters for each block and both experimental halves. Mean amplitudes of the ERP were calculated for electrodes Fz, Cz, Pz and Oz of the 10±20 system (Jasper,1958) for six consecutive time-windows of 100 ms length beginning 250 ms post-stimulus. We decided to use only these four midline electrodes out of the array of 61 recorded because a visual inspection of the data indicated that the obtained e€ects have a broad rather than a focussed scalp distribution. Furthermore, changes in the scalp distribution of the e€ects are not the prime issue of this paper. To test whether motor deviants activated the incorrect (expected) response prior to activation and execution of the correct response, we used t-tests for consecutive time-windows of 10 ms to test whether the LRP showed a signi®cant positivity (LRP ÔdipÕ). The same was done for standards and perceptual deviants to see whether an early activation of the correct response was present (early negativity in the LRP). The ANOVA approach was used to analyze the repeated measure design. Separate analyses were run for error, RT, ERP and LRP data. For the analysis of ERPs, we ®rst formulated a global ANOVA. Subsequent analyses were run according to signi®cant interactions (see results section for details). In all ANOVAs, the degrees of freedom were adjusted to control for violations of the sphericity-assumption where appropriate (Huynh & Feldt, 1980). Degrees of freedom are reported before, pvalues after the adjustment.

54

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

3. Results 3.1. Behavioral data Post-experimental questionnaire. Based on the post-experimental questionnaire two groups of subjects were formed according to the following criteria: Subjects with a recognition score of 7 or a free-recall rate below 38% were categorized as implicit learners (see Eimer et al., 1996, who used similar criteria). These subjects have no or only a small amount of explicit sequence knowledge as assessed by our free recall and recognition procedures. Ten subjects were categorized as implicit, the remaining nine as explicit. Table 1 shows the recognition and free-recall scores for all subjects. The two scores correlate with r ˆ 0:85…P < 0:0001† and thus seem to tap on the same explicit knowledge base. Therefore, it seems to be justi®ed to use a combination of the two measures for categorization 1. Errors. Overall error-rate was small (5.08%) and did not di€er between explicit and implicit learners. Error data were analyzed by means of a three-way repeated measures ANOVA with GROUP (explicit (E) vs. implicit (I)) as between and experimental HALF (1 vs. 2) and STIMULUS TYPE (standard (std) vs. pseudorandom vs. perceptual deviants (pd) vs. motor deviants(md)) as within-subject factors. The interaction HALF by STIMULUS TYPE was signi®cant …F …3; 51† ˆ 3:67; P < 0:018; e ˆ 1.0987†. Inspection of Fig. 2 and post-hoc contrasts reveal the following pattern: In the ®rst half of the experiment, the error rate did not di€er signi®cantly for any of the four stimulus types. In the second half, the error rate remained about the same size for random sequences, standards and perceptual deviants, but increased signi®cantly for motor deviants. Reaction time. To assess learning of the sequential structure of the stimulus material, we compared the mean RT of the pseudorandom blocks (20, 36) with the mean RT of the preceding and the following regular blocks (19, 21, 35 and 37). Mean RT of standard and pseudorandom stimuli for these blocks were submitted to a twoway repeated measures ANOVA with BLOCK (6) as repeated measures and GROUP (E vs. I) as between-subjects factor. Mean RT in pseudorandom blocks was prolonged as compared to the preceding and the following regular blocks (main e€ect BLOCK, F(5, 85) ˆ 10.44, P < 0.0001, e ˆ 0.5075; see Fig. 3). Moreover, explicit subjects were disturbed more than implicit subjects in the pseudorandom blocks as revealed by a signi®cant BLOCK by GROUP interaction, (F …5; 85† ˆ 3:12; P < 0:0432; e ˆ 0.5075). Separate paired t-tests for explicit and implicit subjects showed signi®cant learning-e€ects for both groups (see Table 2).

1

Seventeen of the 19 subjects stated that they had noticed sequential regularities (question 1). However, we did not use the answer to this question as a categorization criterion as it is known to be subject to a response bias. Willingham et al. (1993) found that 24.4% of a group of subjects confronted with unstructured material reported to have noticed sequential regularities. Data of questions 2 and 3 were not analyzed due to problems with categorizing the heterogenous answers of the subjects.

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

55

Table 1

Recognition score (R-score), percent of correct items in free recall and group categorization (group: e ˆ explicit, i ˆ implicit) for each subjecta Subject no.

R-score

Free recall

Group

11 7 13 18 12 2 17 5 19 14 15 4 1 16 6 9 3 10 8

)2 0 0 0 0 2 3 4 7 13 12 13 15 17 20 22 29 42 44

0 25 0 0 25 0 0 37.5 25 25 100 37.5 50 37.5 75 100 100 100 100

i i i i i i i i i i e e e e e e e e e

a Note: Recognition scores were computed as the sum of scores for recognition of bigrams, trigrams and quadrupels and could vary between ‹60. Negative scores or 0 indicate that the subject did not possess any explicit sequence knowledge, positive scores indicate di€erent degrees of explicit sequence knowledge. Free-recall is de®ned by the percentage of correctly recalled consecutive letters. See text for further details.

Fig. 2. Percentage of errors separated for stimulus type (standard letters, pseudorandom letters, perceptual deviants, motor deviants) and experimental half (1 vs. 2).

Visual inspection of Fig. 3 suggests that implicit learners respond faster to standard letters in the ®rst experimental half whereas explicit learners are faster in the second half. This is con®rmed by a GROUP by HALF-interaction in a two-way ANOVA for standard stimuli …F …1; 17† ˆ 4:72; P < 0:0443†.

56

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

Fig. 3. Mean RTs (in ms) of successive blocks for standard letters (blocks 5±19, 21±35, 37±38) and pseudorandom letters (blocks 1±4, 20, 36). Solid line: explicit group, dashed line: implicit group. Table 2

Pairwise comparisons of mean RTs derived from pseudorandom blocks (20, 36) and the preceding and following regular blocks (19, 21 and 35, 37)a Comparison (block)

Group Explicit

20 20 36 36

vs. vs. vs. vs.

19 21 35 37

Implicit

Dt (ms)

t

Dt (ms)

t

254 143 303 277

3.45 3.51 2.61 2.45

62 87 104 148

1.25 2.90 3.63 3.51

a

Note: df ˆ 8 for explicit, df ˆ 9 for implicit group. P < 0.05. ** P < 0.01. *

To delineate the contributions of stimulus and response anticipation to sequence learning, we compared the RTs of standards, perceptual and motor deviants. Separate ANOVAs were run for explicit and implicit subjects with factors STIMULUS TYPE (std vs. pd vs. md) and HALF (2). For explicit as well as implicit learners RTs to standards, perceptual and motor deviants di€ered signi®cantly (main e€ect STIMULUS TYPE, explicit group: F …2; 16† ˆ 8:97, P < 0.01, e ˆ 0.6339; implicit group: F …2; 18† ˆ 21; P < 0.0001, e ˆ 1.08). For explicit subjects, the e€ect of stimulus type became more pronounced in the second half of the experiment (HALF by STIMULUS TYPE interaction, F(2, 16) ˆ 11.94, P < 0.001, e ˆ 0.9165; see Fig. 4). In the second experimental half, planned contrasts for explicit learners revealed highly signi®cant di€erences between perceptual deviants and standards (821 vs. 707 ms, F …1; 8† ˆ 13:43; P < 0:0064† and between motor deviants and perceptual deviants (959 vs. 821 ms, F …1; 8† ˆ 7:6; P < 0:0248†. For implicit learners, only the contrasts `standards vs. motor deviants' (751 vs. 840 ms, F …1; 9† ˆ 21:74; P < 0:0012† and `perceptual vs. motor deviants' (774 vs. 840 ms, F …1; 9† ˆ 5:8;

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

57

Fig. 4. Mean RT (in ms) for regular blocks shown separately for standard letters, perceptual and motor deviants and for both post-experimental groups and experimental halves.

P < 0:039† reached signi®cance in the second experimental half. The RT di€erence between standards and perceptual deviants was not reliable in this group (751 vs. 774 ms, F …1; 9† ˆ 1:57; P < 0:24). 2 Taken together, these results clearly show that implicit and explicit learners respond di€erently to perceptual and motor deviants. While both groups register deviations from the motor sequence, it is only the group of explicit learners who registers deviations from the perceptual sequence, too. Event-related potentials. The most prominent feature of the stimulus-locked ERPs is a positive complex peaking over the parieto±occipital part of the scalp. This positivity starts at about 200 ms after stimulus-onset. It reaches its maximum between 300 and 600 ms and resolves completely at about 800±900 ms (see Fig. 5). Topography and latency of this positivity suggest that it is a member of the P300 family. The rising ¯ank and the maximum of the positivity is interrupted by smaller peaks which are modulated by the experimental variables. A superordinate ANOVA was run with factors GROUP (E vs. I) as between-subjects factor and ELECTRODE (Fz, Cz, Pz, Oz), STIMULUS TYPE (std vs. pd vs. md), HALF (1 vs. 2) and TIME (250±850 ms in six steps of 100 ms) as repeated measures factors. Two signi®cant four-way interactions with factors GROUP showed that the ERPs of explicit and implicit subjects responded di€erently to the experimental manipulations (HALF by ELECTRODE by TIME by GROUP: F …15; 255† ˆ 3:08; P < 0:0197; e ˆ 0.2799) and STIMULUS TYPE by ELECTRODE by TIME by GROUP (F …30; 510† ˆ 3:19; P < 0:0002; e ˆ 0.4379). Therefore, the data of the two groups were analyzed separately. Moreover, the fact that the experimental factors STIMULUS TYPE, HALF, ELECTRODE and GROUP interacted with factor TIME justi®ed separate analyses for di€erent time-windows.

2

We obtained the same results when only data of blocks 31±35 were analyzed. Thus, for implicit learners, there was no signi®cant sensitivity to the perceptual deviance of letters at the end of the experiment.

58

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

Fig. 5. ERPs to standards, perceptual and motor deviants at midline electrodes Fz, Cz, Pz and Oz are depicted separately for the two groups and each experimental half. The time-window extends from 200 ms pre-to 1000 ms post-stimulus. Note the enhanced negativity for deviant letters between 250 and 350 ms post-stimulus (N200 e€ect) and the enhanced P300 amplitude for motor deviants in the second experimental half in the explicit group (left panel).

3.2. Explicit group N200 amplitude. Perceptual and motor deviants evoked a larger negative peak than standards between 250 and 350 ms after stimulus-onset (see Fig. 5). This was con®rmed by a signi®cant main e€ect of STIMULUS TYPE (F …2; 16† ˆ 17:67; P < 0:0004; e ˆ 0.7722) in a three-way repeated measures ANOVA with factors HALF, STIMULUS TYPE (std, pd, md) and ELECTRODE (Fz, Cz, Pz, Oz). A marginally signi®cant STIMULUS TYPE by ELECTRODE-interaction (F …6; 48† ˆ 2:52; P < 0:0642; e ˆ 0.6345) indicated that the N200 amplitude e€ect was larger at central and parietal midline electrodes (Cz: std: 3.6 lV, pd: 0.68 lV, md: 0.74 lV; note that the di€erence between pd and md is not signi®cant). P300 amplitude. Motor deviants evoked a larger P300 amplitude (350±650 ms poststimulus) than standards and perceptual deviants (see Fig. 5). This e€ect was more pronounced in the second half of the experiment and had its maximum at central and parietal electrodes (STIMULUS TYPE by HALF by ELECTRODEinteraction in a three-way repeated measures ANOVA, F …6; 48† ˆ 4:24; P < 0:0053; e ˆ 0.7386). 3.3. Implicit group N200 amplitude. A three-way ANOVA with factors HALF, STIMULUS TYPE and ELECTRODE revealed a signi®cant main e€ect HALF (F(1, 9) ˆ 26.65, P < 0.0006) and a signi®cant interaction HALF by ELECTRODE (F(3, 27) ˆ 5.47, P < 0.0197, e ˆ 0.5663), but neither the main e€ect nor any interaction of factor

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

59

STIMULUS TYPE reached signi®cance, i.e. the N200 amplitude was the same for all three stimulus types. P300 amplitude. Amplitudes for standards, perceptual and motor deviants did not di€er in the P300 time-window 350±650 ms poststimulus. In a three-way repeated measures ANOVA only the STIMULUS TYPE by ELECTRODE-interaction reached signi®cance …F …6; 54† ˆ 4:4; P < 0:0051; e ˆ 0.6774). To summarize, explicit learners showed an enhanced N200 amplitude for both, perceptual and motor deviants and an enhanced P300 amplitude for motor deviants only. Neither the one nor the other e€ect was found for implicit learners. Stimulus-locked LRPs did not di€er for explicit and implicit learners. This was con®rmed by a four-way repeated measures ANOVA with GROUP (E vs. I) as between and HALF (1 vs. 2), STIMULUS TYPE (std vs. p.d, vs. md) and TIME (mean LRP amplitude in 20 consecutive 50 ms time-windows starting with stimulusonset) as within-subjects factors. None of the interactions with factor GROUP reached signi®cance (smallest P > 0.22), nor was the main e€ect GROUP …F …1; 17† < 1† reliable. Therefore, we pooled the LRP data of implicit and explicit learners for further analyses. The main ®nding is summarized in Fig. 6. In the ®rst half of the experiment, all three LRPs are virtually aligned. In the second half of the experiment, however, the LRP which precedes responses to motor deviants di€ers from the other two. It shows a clear positive dip before it starts to rise in the negative direction. This suggests that the incorrect hand was partially activated before the response with the correct hand was executed. In contrast, the LRP for standards and perceptual deviants shows an early negativity which indicates activation of the correct response. The positive dip for motor deviants had an amplitude of 0.47 lV between 70 and 80 ms poststimulus. A t-test against 0 provided t(1, 18) ˆ 2.11, P < 0.0482). The positive-going trend started to develop already before stimulus-onset, i.e. the next response was already predicted by the system on the basis of the previous response without a full analysis of the currently presented stimulus. Standards and perceptual deviants gave rise to an early negativity in the LRP (®rst time-frame with a negative LRP reliable di€erent

Fig. 6. Grand average of the stimulus-locked lateralized readiness potentials. The LRP is shown separately for standards, perceptual and motor deviants, and the ®rst vs. second experimental half starting 500 ms pre- and ending 1000 ms post-stimulus. Negative amplitudes indicate activation of the correct, positive amplitudes indicate activation of the incorrect response.

60

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

from 0: standards: 30±40 ms …ÿ0:23 lV; t…1; 18† ˆ 2:279; P < 0:0351†; perceptual deviants: )30 to )20 ms )0.27 lV; t…1; 18† ˆ 2:14; P < 0:0464†† indicating activation of the correct response. Additional analyses. To test whether group di€erences are present in the late part of P300 resolution which might be suggested by Fig. 5, we computed an additional ANOVA in the time-frame 800±1000 ms after stimulus-onset (GROUP (E vs. I) by TYPE (std, pd, md) by ELECTRODE (Fz, Cz, Pz, Oz) by HALF (1 vs. 2)). No main e€ect GROUP emerged in this analysis. Furthermore, no interaction involving the factor GROUP reached signi®cance. Due to the experimental design, the probability of an immediate response repetition is 12.5% for standards and perceptual deviants and 25% for motor deviants. Thus, the two factors `response repetition probability' and `type of stimulus' were partially confounded and the ERP and RT di€erences observed between the different stimulus types could also be due, at least in part, to the di€erent probabilities of a response repetition. To test for this possible e€ect, we replicated all statistical analyses for RT, ERP and LRP measures but excluded response repetition trials. These analyses provided the same pattern of results as outlined above. Thus, the argument that the e€ects are partially caused by the di€erent response repetition probabilities has to be rejected. 4. Discussion In a four-choice RT task two letters each were related to one response, and two types of deviants occasionally replaced regular letters in an otherwise regular sequence of events. Perceptual deviants changed the stimulus- but not the responsesequence whereas motor deviants changed the stimulus- as well as the responsesequence. This enabled us to disentangle stimulus- and response-related sequence learning. Subjects were categorized as those who possessed and those who did not possess explicit, i.e. verbalizable knowledge about the previously presented sequence on the basis of a post-experimental test. Both groups of subjects learned the underlying sequential regularities despite the fact that in each string of eight events one regular element was always replaced by a deviating letter. This is re¯ected by a prolongation of the average response time in unstructured (i.e. pseudorandom) compared to structured stimulus blocks, and by an overall decrease of response time for standard letters during the total course of the experiment. Learning is also re¯ected by a higher error rate for motor deviants in the second half of the experiment, i.e. as the subjects have acquired more knowledge about the regularities of the event sequence they are more disturbed by events which require an unexpected response. These ®ndings are in line with the results of other investigators who, too, reported learning of sequential regularities, despite the fact that these were interrupted from time to time by unpredictable elements (Cleeremans & McClelland, 1991; Eimer et al. 1996, Jimenez, Mendez, Cleeremans, 1996; Stadler, 1993).

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

61

As found in previous studies, subjects with explicit knowledge showed a larger performance gain than subjects with implicit knowledge (e.g. Curran & Keele, 1993; Mayr, 1996). In our study, this performance advantage developed gradually after block 24. In the ®nal blocks, the average response time of implicit learners was about 100 ms longer than that of explicit learners. However, the opposite held in the ®rst part of the experiment. Between blocks 9 and 18 explicit learners were about 100 ms slower than implicit learners. This pattern of results suggests the following interpretation: Explicit learners might have noticed regularities in the sequence rather early. From that time onward they might have tried to ®gure out the exact sequential rules and, therefore, might have acted as in a dual task situation ± the primary task being the choice reaction time task and the secondary comprising the extraction of the sequential rules. The secondary task might have captured some of the available processing resources and this resource trade-o€ might be re¯ected by the prolongation of the choice RTs. Later on, the explicit learners might have acquired complete knowledge about the event sequence and could most likely predict the next event much more e€ectively than implicit learners. The response times to the di€erent types of stimuli also show that the two groups of subjects had acquired a di€erent amount and a di€erent quality of sequence knowledge. Performance of implicit learners was impaired by motor deviants only. These stimuli prolonged the average response time for about 50 ms in the ®rst and for about 100 ms in the second half of the experiment. This suggests that implicit learners must have acquired some knowledge about the sequence of motor responses in the ®rst half of the experiment already. This knowledge was strengthened in the second half but overall the improvement was limited. Explicit learners seem to have been disturbed by both, perceptual and by motor deviants and in both cases the prolongation of response time was more pronounced than in the group of implicit learners. In the ®rst half of the experiment, motor deviants and standards di€ered, as in the group of implicit learners, for about 50 ms only. However, in the second half, the response time di€erence increased dramatically and amounted to about 250 ms. This e€ect was caused by two changes: a very pronounced decrease of the RT to standards and a slight increase of RT to motor deviants. Perceptual deviants were in between these two extremes, i.e. responses to perceptual deviants were about 25 ms slower in the ®rst half and 125 ms slower in the second half of the experiment. All this suggests that subjects with explicit knowledge could predict the next response extremely well but they also had to pay for these extrapolations. Any deviation from the expected event sequence impeded their response, and these costs were larger for motor than for perceptual deviants. This implies that subjects with explicit knowledge must have represented R±R or S±R dependencies as well as S±S dependencies. Note that all of these changes of response times cannot be attributed to a speed-accuracy trade-o€. Error rates were very small and did hardly di€er between conditions, and in the only case in which error rate increased (i.e. for motor deviants in the second half of the experiment) this increase coincided with an increase of response time. It is surprising that implicit learners acquired knowledge of the response dependencies only although the response sequence was more complex than the stimulus

62

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

sequence (see methods). This failure to extract the regularities of the `directly visible' stimulus sequence suggests that these subjects did not pay attention at all to the repeating sequence of letters. They performed the choice reaction time task as such and acquired knowledge of the response dependencies just by doing, not by `thinking about regularities'. This, too, supports the notion that the two groups of subjects seem to have handled the task in a di€erent manner. In the group of implicit learners both N200 and P300 components of the ERP were evoked with the same latency and amplitude by standard letters, perceptual, and motor deviants. Since both components are assumed to re¯ect stimulus evaluation processes, this ®nding suggests that all three types of stimuli were processed by the system in the very same manner. Obviously, the brain of implicit learners made no distinction between these three stimuli, or, in other words, the fact that the otherwise regular sequence was interrupted by a di€erent letter was not recognized at all by those processors which evaluate the perceptual input and which perform an update of the internal model of the environment. In the group of explicit learners, however, both types of deviants evoked a much more pronounced N200 than standards in the second half of the experiment, i.e. when explicit knowledge about the stimulus sequence had accumulated. The N200 e€ect indicates that any perceptual inconcistency in the stimulus sequence was noticed by the system. A comparable N200 e€ect for motor deviants was described by Eimer et al. (1996) for subjects categorized as explicit learners. These authors suggested that the N200 e€ect could re¯ect the amount of consciously available sequence knowledge. This conclusion rests on the assumption that the tests for explicit knowledge are process pure measures, i.e. not contaminated by implicit knowledge (Shanks & St. John, 1994). This is a strong claim which may be dicult to substantiate. However, even if the knowledge tests are not process pure measures and if the N200 e€ect cannot be functionally related to the subjective domain of awareness the data nevertheless provide converging evidence for a weaker conclusion, namely that a memory trace of the stimulus sequence could have been stored. The N200 e€ect for perceptual deviants in our study suggests that this memory trace could represent stimulus attributes as such without being contaminated with motor representations. Beyond this, the ERPs of explicit learners revealed another systematic e€ect: The amplitude of the P300 was reliably larger for motor deviants than for perceptual deviants or standards. This e€ect, too, popped out in the second half of the experiment, i.e. when knowledge about the event sequence was clearly available. This ®nding replicates and extends the ®ndings of Eimer et al. (1996) who obtained a slight enhancement of the P300 for motor deviants. The ERPs to perceptual deviants in the present experiment provide further information with respect to the P300 e€ect. Since the N200 was a€ected by both types of deviants and the P300 by motor deviants only, one has to conclude that the generating mechanisms are functionally distinct. Assuming that the system continuously generates and updates a model which comprises all perceptual features of the next stimulus to be expected, Gehring et al. (1992) argued that the N200 component re¯ects a process which is sensitive to any deviation of an actually perceived stimulus from such a

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

63

model. In our study both perceptual and motor deviants di€ered perceptually from the next most likely, regular stimulus. Thus, it makes sense that both bear an e€ect on the N200. The motor deviants, on the other hand, formed not only a mismatch with an expected stimulus template, but, in addition, transmitted task relevant information, because after perceiving it the subject had to change a primed action or motor program. Task relevance in this shade of meaning has been found to be one of the most potent antecedent conditions for the P300 component (Donchin & Coles, 1988; Johnson, 1986). Among others, the amplitude of P300 proved to be larger for stimuli which required an overt behavioral response than just a silent counting response, and the P300 amplitude was also found to be larger in case of a silent counting response than if a rare stimulus had only be watched passively. In our study, encountering a motor deviant did not only enforce an update of the stimulus sequence model but also an update of the currently held action model. It seems likely that this additional updating step is re¯ected by the increase of P300 amplitude. The LRP provides further insight into the mechanisms of sequence learning. As mentioned above, the polarity of the stimulus-locked LRP reveals whether the centrally initiated motor program is correct or incorrect. With respect to the LRP both groups showed the same response pattern. In the second experimental half the LRP accompanying motor deviants was clearly di€erent from that accompanying standards or perceptual deviants: The initial part of the LRP for motor deviants was positive (see also Eimer et al., 1996). This suggests that the program for the expected, but inappropriate hand was at least partially activated (Coles, 1989; Gratton et al., 1988). Most likely, this is an automatic process which has to be inhibited after a full analysis of the stimulus (Gratton et al., 1988). The ®nding that both groups show an initial activation of the incorrect hand suggests that priming of the next response depended in both groups on the same type of response-to-response association. The LRP to perceptual deviants (which were not present in the Eimer et al. (1996) studies) is very similar to that of standards, i.e. it shows an early activation of the correct response in the second half of the experiment. Thus, it provides further information on the type of representation subjects acquire in the SRT task which go beyond those in the Eimer et al. (1996) experiment. On a motoric level, perceptual deviants are processed in the same way as standard letters. However, for explicit learners, evaluation of the stimulus seems to lead to a mismatch between the expected letter and the one actually presented which is re¯ected in the N200 e€ect. Taken together, the observed pattern of RT and ERP results is compatible with the idea that implicit learners represent mainly, if not solely, R±R dependencies while explicit learners represent both S±S and R±R dependencies. This seems to hold, at least, for event sequences in which symbols and not spatial positions are used as stimuli (see Nattkemper & Prinz, 1997). These results converge with recent PET studies in which cerebral blood ¯ow increased substantially during implicit sequence learning in the motor areas contralateral to the performing hand while not much

64

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

blood ¯ow change could be observed in the sensory association areas (Doyon et al., 1996; Grafton et al., 1995; Hazeltine et al., 1997).

Acknowledgements This work was supported by the German Research Foundation (DFG, grant Ro 529) and the Berlin-Brandenburg Academy of Sciences (BBAW). We thank Kerstin  Jost, Mustafa Oczan and Bettina Rolke for help during data acquisition, Hansjerg G olz for programming support and Michael Stadler, Martin Eimer, Axel Cleeremans and two anonymous reviewers for helpful comments on previous versions of the manuscript. Appendix A List of bigrams, trigrams and quadrupels used for the recognition procedure. Regular bigrams, trigrams and quadrupels which were part of the sequence: TX VLK VLKT KT TXS KTXS XS SMR TXSM MR KTX XSMR SM RVL LKTX Irregular bigrams, trigrams and quadrupels which were not part of the sequence: LT VSL VTXM VS KSR VLTR KX LKS LKXM MV SMX TXSR SK XSR KTLV References Cherry, K. E., & Stadler, M. A. (1995). Implicit learning of a nonverbal sequence in younger and older adults. Psychology and Aging, 10 (3), 379±394. Clegg, B. A., DiGirolamo, G. J., & Keele, S. W. (1998). Sequence learning. Trends in Cognitive Sciences, 2 (8), 275±281. Cleeremans, A., & McClelland, J. L. (1991). Learning the structure of event sequences. Journal of Experimental Psychology, 120 (3), 235±253. Cohen, A., Ivry, R. I., & Keele, S. W. (1990). Attention and structure in sequence learning. Journal of Experimental Psychology: Learning Memory and Cognition, 16 (1), 17±30. Coles, M. G. H. (1989). Modern mind brain reading: Psychophysiology, physiology and cognition. Psychophysiology, 26 (3), 251±269. Courchesne, E., Courchesne, Y., & Hillyard, S. A. (1978). The e€ect of stimulus deviation on P3 waves to easily recognized stimuli. Neuropsychologia, 16, 189±199. Curran, T., & Keele, S. W. (1993). Attentional and nonattentional forms of sequence learning. Journal of Experimental Psychology: Learning, Memory and Cognition, 19 (1), 189±202.

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

65

Curran, T. (1998). Implicit sequence learning from a cognitive neuroscience perspective: What, how and where? In: M.A. Stadler, & P.A. Frensch, Handbook of implicit learning (pp. 365±400). London: Sage. De Jong, R., Wierda, M., Mulder, G., & Mulder, L. J. M. (1988). Use of partial stimulus information in response processing. Journal of Experimental Psychology: Human Perception and Performance, 14, 682±692. Donchin, E., & Coles, M. G. H (1988). Is the P300-component a manifestation of context updating?. Behavioral and Brain Sciences, 11, 355±372. Doyon, J., Owen, A. M., Petrides, M., Sziklas, V., & Evans, A. C. (1996). Functional anatomy of visuomotor skill learning in human subjects examined with positron emission tomography. European Journal of Neuroscience, 8, 637±648. Duncan-Johnson, C., & Donchin, E. (1982). The P300 component of the event-related brain potential as an index of information-processing. Biological Psychology, 14, 1±52. Eimer, M., Goschke, T., Schlaghecken, F., & St urmer, B. (1996). Explicit and implicit learning of event sequences: evidence from event-related brain potentials. Journal of Experimental Psychology: Learning, Memory and Cognition, 22 (4), 970±987. Frensch, P. A., & Miner, C. S. (1994). E€ects of presentation rate and individual di€erences in short-term memory capacity on an indirect measure of serial learning. Memory and Cognition, 22 (1), 95±110. Gehring, W. J., Gratton, G., Coles, M. G. H., & Donchin, E. (1992). Probability e€ects on stimulus evaluation and response processes. Journal of Experimental Psychology: Human Perception and Performance, 18 (1), 198±216. Goschke, T. (1998). Implicit learning of perceptual and motor sequences: Evidence for independent learning systems. In: M.A. Stadler, P.A. Frensch, Handbook of implicit learning (pp. 401±444). London: Sage. Grafton, S. T., Hazeltine, E., & Ivry, R. (1995). Functional mapping of sequence learning in normal humans. Journal of Cognitive Neuroscience, 7, 497±510. Gratton, G., Bosco, C. M., Kramer, A. F., Coles, M. G. H., Wickens, C. D., & Donchin, E. (1990). Eventrelated brain potentials as indices of information extraction and response priming. Electroencephalography and Clinical Neurophysiology, 75, 419±432. Gratton, G., Coles, M. G. H., Sirevaag, E. J., Eriksen, C. W., & Donchin, E. (1988). Pre- and poststimulus activation of response channels: A psychophysiological analysis. Journal of Experimental Psychology: Human Perception and Performance, 14, 331±344. Hackley, S. A., & Miller, J. (1995). Response-complexity and precue interval e€ects on the lateralized readiness potential. Psychophysiology, 32, 230±241. Hazeltine, E., Grafton, S. T., & Ivry, R. (1997). Attention and stimulus characteristics determine the locus of motor sequence encoding: a PET study. Brain, 20, 123±140. Ho€mann, J., & Koch, I. (1997). Stimulus±response compatibility and sequential learning in the serial reaction time task. Psychological Research, 60, 87±97. Howard, D. V., & Howard, Jr., J. H. (1989). Age di€erences in learning serial patterns: direct vs. indirect measures. Psychology and Aging, 4 (3), 357±364. Howard, D. V., & Howard, Jr., J. H. (1992). Adult age di€erences in the rate of learning serial patterns: evidence from direct and indirect tests. Psychology and Aging, 7 (2), 232±241. Howard, Jr., J. H., Mutter, S. A., & Howard, D. V. (1992). Serial pattern learning by event observation. Journal of Experimental Psychology: Learning, Memory and Cognition, 18 (5), 1029±1039. Huynh, H., & Feldt, L. A. (1980). Conditions under which mean square rations in repeated measurement designs have exact F-distributions. Journal of the American Statistical Association, 65, 1582±1589. Jasper, H. H. (1958). The ten±twenty-electrode-system of the international federation. Electroencephalography and Clinical Neurophysiology, 20, 371±375. Jimenez, L., Mendez, C., & Cleeremans, A. (1996). Comparing direct and indirect measures of sequence learning. Journal of Experimental Psychology: Learning, Memory and Cognition, 22 (4), 948±969. Johnson, R. (1986). A triarchic model of the P300 amplitude. Psychophysiology, 23, 367±384. Johnson, R.S. (1988). The amplitude of the P300 component of the event-related potential: review and synthesis. In: P.H. Ackles, J.R. Jennings, M.G.H. Coles, Advances in Psychophysiology, vol. 3 (pp. 62± 138). Greenwich, CT: JAI press.

66

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

Keele, S. W., Jennings, P., Jones, P., Caulton, D., & Cohen, A. (1995). On the modularity of sequence representation. Journal of Motor Behavior, 27, 17±30. Kornhuber, H. H., & Deecke, L. (1965). Hirnpotential anderungen bei willk urbewegungen und passiven bewegungen des menschen: bereitschaftspotential and rea€erente potentiale. P¯ ugers Archiv F ur Die Gesamte Physiologie, 248, 1±17. Kutas, M., & Donchin, E. (1980). Preparation to respond as manifested by movement related brain potentials. Brain Research, 202, 95±115. Matt, J., Leuthold, H., & Sommer, W. (1992). Di€erential e€ects of voluntary expectancies on reaction times and event-related potentials: evidence for automatic and controlled expectancies. Journal of Experimental Psychology: Learning, Memory and Cognition, 18 (4), 810±822. Mayr, U. (1996). Spatial attention and implicit sequence learning: evidence for independent learning of spatial and nonspatial sequences. Journal of Experimental Psychology: Learning, Memory and Cognition, 22 (2), 350±364. Miller, J., & Hackley, S. A. (1992). Electrophysiological evidence for temporal overlap among contingent mental processes. Journal of Experimental Psychology, 121, 195±209. Nattkemper, D., & Prinz, W. (1997). Stimulus and response anticipation in a serial reaction task. Psychological Research: Sequence learning (special issue). Phenomena and Models, 98±112. Nissen, M. J., & Bullemer, P. (1987). Attentional requirements of learning: evidence from performance measures. Cognitive Psychology, 19, 1±32. Osman, A., Bashore, T., Coles, M. G. H., Donchin, E., & Meyer, D. E. (1992). On the transmission of partial information: inferences from movement-related brain potentials. Journal of Experimental Psychology: Human Perception and Performance, 18 (1), 217±232. Osman, A., & Moore, C. M. (1993). The locus of dual-task interference: psychological refractory e€ects on movement-related brain potentials. Journal of Experimental Psychology: Human Perception and Performance, 19 (6), 1292±1312. Osman, A., Moore, C. M., & Ulrich, R. (1995). Bisecting RT with lateralized readiness potentials: precue e€ects after LRP onset. Acta Psychologica, 90, 111±127. Reed, J., & Johnson, P. (1994). Assessing implicit learning with indirect tests: determining what is learned about sequence structure. Journal of Experimental Psychology: Learning, Memory and Cognition, 20 (3), 585±594. Ritter, W., Ford, J. M., Gaillard, A. K. W., Harter, M. R., Kutas, M., N a at anen, R., Polich, J., Renault, B., & Rohrbaugh, J. (1984). Cognition and event-related potentials: 1. The relationship of negative potentials and cognitive processes. Annals of the New York Academy of Science, 425, 24±38. Ritter, W., Simson, R., & Vaughan, Jr., H. G. (1983). Event-related potential correlates of two stages of information processing in physical and semantic discrimination tasks. Psychophysiology, 20 (2), 168±179. Shanks, D. R., & St. John, M. F. (1994). Characteristics of dissociable human learning systems. Behavioral and Brain Sciences, 17, 367±447. Sommer, W., Leuthold, H., & Ulrich, R. (1994). The lateralized readiness potential preceding brief isometric force pulses of di€erent peak force and rate of force production. Psychophysiology, 31, 503±512. Sommer, W., Matt, J., & Leuthold, H. (1990). Consciousness of attention and expectancy as re¯ected in event-related potentials and reaction-times. Journal of Experimental Psychology: Learning, Memory and Cognition, 16 (5), 902±915. Squires, K. C., Donchin, E., Herning, R. I., & Mc Carthy, G. (1977). On the in¯uence of task relevance and stimulus probability on event-related potential components. Electroencephalography and Clinical Neurophysiology, 42, 1±14. Stadler, M. A. (1989). On learning complex procedural knowledge. Journal of Experimental Psychology: Learning, Memory and Cognition, 15 (6), 1061±1069. Stadler, M. A. (1992). Statistical structure and implicit serial learning. Journal of Experimental Psychology: Learning, Memory and Cognition, 18 (2), 318±327. Stadler, M. A. (1993). Implicit serial learning: questions inspired by Hebb (1961). Memory and Cognition, 21 (6), 819±827.

J. R usseler, F. R osler / Acta Psychologica 104 (2000) 45±67

67

Stadler, M. A. (1995). Role of attention in implicit learning. Journal of Experimental Psychology: Learning, Memory and Cognition, 21 (3), 674±685. Willingham, D. B., Nissen, M. J., & Bullemer, P. (1989). On the development of procedural knowledge. Journal of Experimental Psychology: Learning, Memory and Cognition, 15 (6), 1047±1060. Willingham, D. B., Greeley, T., & Bardone, A. M. (1993). Dissociation in a serial response time task using a recognition measure: Comment on Perruchet and Amorim. Journal of Experimental Psychology: Learning, Memory and Cognition, 19 (6), 1424±1430.