Testing for a dissociation: comparing practice, observation and non­ practice groups on the serial reaction time task and on a free-generation task

with a six item unique sequence.

According to Cohen, Ivry & Keele (1990) and Berry (1991) conditions for observational learning may be maximised by utilising sequences which are deemed unique in their inherent structure and where the rule governing the order of location of sequence elements is salient. Experiment 4.2 will utilise the six item unique sequence employed in Experiment 4.1 where each element in the sequence provides a unique marker for the position of the next element within the sequence and where sixteen sequence exposures per block of trials is employed. In addition Experiment 4.2 will include a greater number of participants per group in order to maximise the chances of observing a differential observational learning effect when compared with a non-practice group.

Method

Participants

Twenty-seven participants took part in the study ( 14 females, 13 males). All were students at the Department of Psychology, U.C.L. Ages ranged from 18 yrs to 35 yrs. All reported normal or corrected vision.

Apparatus & Materials

same order and manner as they were employed in Experiment 4.1. Participants, firstly took part in the serial reaction time task followed after two further blocks of refresher blocks by the free generation task.

Three groups of subjects were employed in the study; practice (n=9) , observation (n=9) and non-practice (n=9). Subjects carried out the SRT task and the free generation task as outlined in Experiment 4.1. The experimenter took the place as demonstrator for participants in the observation condition. Demonstrator training on the new sequences was carried out as before.

Procedure

Participants were tested in a manner identical to the 16 exposure condition in Experiment 4.1.

Results

Serial reaction time task

550 T 500 Practice Observation Non Practice 450 400 350 300 O) Block number

F ig u re 4 .2 . Graph showing m ean R Ts (ms) for each block of training and testing for practice, observation and non-practice on the serial reaction tim e task employing a unique six-item

sequence.

Figure 4.2 shows mean RT for each block of training and each block of testing by group on the serial reaction time task. The means showed that practice subjects produced a large elevation in RT between block 8 and block 9 implying a large amount of interference to RT caused by the introduction of the alternative sequence in block 9 of the task. In contrast, observation and non-practice subjects appeared to show equivalent elevation in RT between block 8 and block 9 albeit that RT levels for observers were lower at block 8 in comparison to the non-practice group.

These impressions were confirmed following a three-way ANOVA (block x group

X counterbalancing) which yielded a significant main effect of group E(2,21) = 4.94,

p<0.05, a significant main effect of block F(1,21) = 111.81, p<0.05, a reliable group by block interaction F(2,21) = 9.76, p<0.05, a reliable counterbalancing by block interaction F(1,21) = 5.57, p<0.05, but no three way interaction between group x block X counterbalancing F<1. Analysis of the interaction between counterbalancing

condition and block revealed that, as in Experiment 3.2, greater RT elevation occurred between block 8 and block 9 when the training sequence contained a salient chunk, in this case, 1234. The effect of counterbalancing was uniform across groups.

Following up the interaction of block x group, analysis of the simple effects of block at each level of group, practice, observation and non-practice all showed a reliable, but not equal, elevation in RT between block 8 and block 9, smallest F(1,21) = 15.55, p<0.05. RT elevation was greatest for the practice group than for observation and non-practice which were similar. Comparing each group's performance on the SRT task at each block in turn, at block 8 Tukeys follow up tests revealed that the practice group produced faster RT's than the observation group which produced faster RT's than the non-practice group. Tukeys tests at block 9 found that the practice group produced faster RT's than non-practice, and the observation group

produced faster RT's than both the practice and non-practice groups. This confirmed that practice was more effective than non-practice and observation in promoting sequence learning, whilst observation did not differ from non-practice.

No group differences were apparent at block 1 of training F<1. A two way ANOVA comparing RT saving between block 1 and block 8 of the SRT task yielded a main effect of block F(1, 25) = 7.51, p<0.05, a reliable group x block interaction E(2,25) = 3.9, p<0.05 but no significant main effect of group F>1. Analysis of the interaction of group by block revealed that both the practice and the observation group showed reliable RT saving between block 1 and block 8, smallest F =10.2. Thus, although this implies evidence of learning on the part of observers, it can only be concluded, on the basis of the similarity of RT elevation between block 8 and block 9 with non­ practice group performance, that observers merely acquired knowledge of general task demands. Error data 4.5 _ 4

1

3.5 d> o> 2 S u o a c ro 0) 3 2.5 2 - 1.5 1 0.5 0 Practice Observation Group Non-Practice

F ig u re 4 .2 a Graph showing the interaction block by exposure for experim ent 4.1 on the SRT task.

A three way ANOVA (group x block x counterbalancing ) of the errors yielded a significant main effect of block F(1,21) = 4.94, p<0.05 but no main effect of group,

counterbalancing or interactions between these factors. Due to the fact that there are only two levels of the factor, block, it was concluded that regardless of group allocation there was a reliable increase in the number of errors made at block 9 (see figure 4.2a)

Free generation task

Mean correct triplets by group M 70 0) Q. 60 ■= 50

t

40 o " 30 o O 20 m 10 0) E 0 Practice Observation Group Non-practice

Fig u re 4 .2b . Graph showing the mean number of triplets which correctly adhered to the SEQ training sequence for practice, observation and non practice groups on the free generation task with

a new six item unique sequence. Error bars represent SEM.

Here, the observation group managed to replicate a mean of 56 triplets which conformed to the sequence on which they were trained compared with practice group subjects who replicated a group mean of 55.11 triplets. 'Correct' scores for these two groups contrasted sharply with non-practice group performance where a group mean of 17.78 correct triplets is reported. From the means it appears that both the observation group and the practice group performed at a similar level.

These impressions were confirmed following a Three-way ANOVA (fg x group x counterbalancing condition) which yielded a significant main effect of group F(2,26) = 4.4, p<0.05, but no effect of counterbalancing and no interaction between the two

factors E>1 • A Fishers LSD test on the group differences revealed that practice and observation did not differ in the number of correct triplets they generated but both practice and observation differed from non-practice. The fact that observers reliably outperformed non-practice participants on this measure suggests that declarative sequence knowledge was apparent under observational conditions.

Discussion

Experiment 4.2 employed a change to the original SOC sequence making the new six item sequence unique in structure, i.e. each dot location signalling only one other dot location, in an attempt to maximise learning conditions for all groups. In addition a free generation task was introduced in order to investigate suggestions that there is a possible dissociation in performance between indirect and direct measures of testing.

Results appeared to support this suggestion. Practice participants outperformed both the observation group and the non-practice group on the S R I task evidenced by a reliable increase in RT between block 8 and block 9. The observation group displayed no evidence of procedural learning on this measure when compared with the non-practice group. However, both the observation group and the practice group were found to reproduce reliably more triplets on the free generation task compared with a non-practice group. Can these findings be regarded as evidence of a true dissociation in performance between indirect and direct measures of testing following observation?

On the free generation task in experiment 4.1 there was no effect of observation. In both cases, participants were exposed to sixteen opportunities to observe the sequence per block of trials. Why is it that an observation effect was forthcoming in experiment 4.2 but not in experiment 4.1 on the free generation task? Firstly, it is difficult to control for observer behaviour during training on the sequence. It is impossible to know how much time a participant devotes to watching the movement

of the dots on the computer screen or the fingers movements of the demonstrator or indeed how much attention to the task they are allocating as a whole. Other factors such as model status, difference in age between demonstrator and subject, sex of the subject have all been found to influence the observational experience, some positively and some negatively (McCullagh 1986). Any subtle changes between the two experiments may have resulted in a confusion of the conditions under which observation learning can take place. The issue of a conflict in attentional resources between observation of the dot movements and the finger movements of the demonstrator will be reserved for a later chapter.

A second suggestion as to why an observation effect occurred here in experiment 4.2 relates to the three blocks of trials (10,11,12) which typically separate the SRT task and the free generation task. Three blocks of differential trails, observation for the observers and anagrams for the non-practice group, may have resulted in observers obtaining all the declarative knowledge they required to outperform non­ practice group participants after they had completed the SRT task. Experiment 4.3 was designed to investigate this possibility. In Experiment 4.3 the refresher blocks were removed in order to assess the effects of their removal on free generation task performance. If, under these circumstances, observation does not generate more correct triplets than non-practice on the free generation task then it would suggest that the declarative learning seen in Experiment 4.2 occurred during these interim blocks. Experiment 4.3 replicated the procedure for observation and non-practice in 4.2 while excluding the differential training blocks between the SRT and free generation task.

In document Observational sequence learning (Page 97-103)