- Email: [email protected]
Stimulus miscuing and dishabituation: electrodermal activity and resource allocation
Biological Psychology 31 (1990) 229-243 0 1990 Elsevier Science Publishers B.V. All rights reserved
229 0301-11511/90/$03..50
STIMULUS MISCUING AND DISHABITUATION: ACTIVITY AND ~SOURCE FLOTATION * David A.T. SIDDLE PACKER
* *, Daphne
BROEKHUIZEN,
ELECTRODERMAL
and Jeanette
S.
School of Behu~~iouralSciences, Macquarie University, New South Wales 2109, Australia
The present research investigated the effects of miscuing a shock stimulus on dishabituation of the skin conductance response and on the allocation of processing resources. In both experiments, a control group received 21 Sl-S2 pairings intermixed with 23 S3-alone presentations. For the experimental group, S2 was miscued on trials 11 and 22 by its presentation following S3. In Experiment 1 (N= 481, S2 was a shock that was either “clearly discernible” or ‘.uncomfortabl~ but not painful”. The results indicated increased electrodermal responding when S2 was miscued by S3 and subsequent dishabituation when S2 again followed Sl on the next trial. Miscuing produced dishabituation with both a high- and low-shock S2. A continuous measure of S2 expectancy revealed that expectancy of S2 in the presence of Sl declined as a result of miscuing. Experiment 2 (N = 36) employed reaction time to a white noise probe stimulus as the dependent variable. The critical data came from probes presented 300 ms following S2 onset on the SI-S2 trial immediately following miscuing. They indicated that miscuing produced slowed reaction time to probes presented during S2 on the following SI-S2 trial. Thus, the miscuing of S2 by S3 appears to result in an increase in processing resources devoted to S2. even when S2 is presented in its usual position following Sl. The results are discussed in terms of current theories of associative learning. Keywords:
Classical conditioning. allocation
Habituation,
Skin conductance,
Associative
learning,
Resource
1. Introduction Some theories of associative learning attempt to account for both habituation and Pavlovian conditioning within the same explanatory framework. For example, Wagner (1978) has argued that one process which underIies re* This research was supported by a grant from the Australian Research Council. The third author was supported by a National Research Fellowship. Thanks are due to Len Glue for assistance with computer programming and to Marie Ayre and Brett Daniels for assistance with data analysis. ** Requests for reprints should be addressed to Dr. David Siddle, Department of Psychology, University of Queensland, Queensland 4072, Australia.
sponse habituation is associatively generated priming. Associatively generated priming occurs as a result of associations developed between events such as contextual cues and the habituation stimulus. Contextual stimuli can then act as retrieval cues to prime into short-term memory (STM) a representation of the habituation stimulus. Similarly, a conditioned stimulus (CS) is said to act as a retrieval cue which primes a representation of the unconditioned stimulus (US) into STM. The extent to which an event is primed in STM is said to determine the degree to which that event is processed. Unprimed events are processed more eIaborately and lead to larger responses than do primed events. A considerable amount of recent data is consistent with an associative account of habituation. Fur example, Siddle and his colleagues have‘ employed a procedure in which paired stimuli (Sl and S2) are presented. They have reported that 52 omission after a number of Sl-S2 pairings results in electrodermal responding at the time of omission and in an increase in responding to a subsequent presentation of S2 when it is represented following Sl (Packer, Siddle, & Tipp, 1989; Siddle, Booth, & Packer. 1987). It can be argued that presentation of Sl-S2 pairings leads to the priming of S2 in STM by Sl. Omission of S2 is thus an unexpected event which is more elaborately processed and which leads to larger electrodermal responses. An SI-alone presentation also weakens the associative strength between Sl and S2 so that the priming of S2 by Sl is Iess effective on the following Sl-S2 trial. Thus, the occurrence of S2 is less expected, and so that event is more elaborately processed and elicits larger responses. There are two types of data consistent with this interpretation. First, response time to secondary task probe stimuli presented during S2 omission and S2 re-presentation is slower than in a no-omission control condition (Siddle & Packer, 1987). To the extent that performance on a secondary task reflects the processing resources allocated to a concurrent primary task (Dawson, Schcll, Beers, & Kelly, 1982), these data indicate that S2 omission and S2 re-presentation each command processing resources. Second, a continuous measure of S2 expectancy indicates that SI-S2 pairings resuIt in a negatively accelerated growth in S2 expectancy in the presence of Sl. In addition, omission of S2 produces a decrease in expectancy of S2 in the presence of Sl (Siddle et al., 1987). Not all data, however, are consistent with an associative analysis. For example, the miscuing of S2 by another event has also been shown to produce dishabituation of the electrodermal response to 52. Siddle (1985) and Packer and Siddle (1989) presented subjects with Sl-S2 pairings intermixed with presentations of S3-alone. On some trials, Sl was replaced by S3 so that S2 was miscued on those trials. Not only did the miscued S2 result in large electrodermal responses, but an S2 presented following Sl on the next trial also elicited larger responses than in a control condition. Moreover,
D.A. T. Siddle et al. / Miscuing and dishabituation
231
reaction time to probes presented during S2 on the miscuing and re-presentation trials was siower than in a control condition. Finally, the expectancy measure indicated that that the miscuing of S2 by S3 (i.e., an S3-S2 presentation) resulted in a decrease in the extent to which S2 was anticipated in the presence of Sl. What these data seem to indicate is that the miscuing of S2 by S3 interfered retrospectively with the associative link between Sl and S2. It is not clear how these results can be accommodated within theories of associative learning in that gains and losses in associative strength are usually said to occur only as a result of reinforcement or non-reinforcement of a CS (e.g., Rescorla & Wagner, 1972). One difficulty, however, is that the stimuli employed in the miscuing studies were tones, lights, and vibrations of moderate intensity. Stimuli of higher intensity and of the type usually employed in studies of Pavlovian conditioning have not been used, and it is possible that the effect of miscuing on dishabituation will not be reproduced when a typical US is employed as S2. The present experiments investigated this possibility. Experiment 1 examined the effects of S2 intensity on responses to the miscued stimulus and on dishabituation of the skin conductance response (SCR). A continuous measure of S2 expectancy was also employed in order to replicate the Packer and Siddle (1989) results which indicated that the miscuing of S2 by S3 reduced expectancy of 52 in the presence of Sl. Experiment 2 examined whether re-presentation following miscuing of an 52 of higher intensity commands processing resources as assessed by reaction time to a secondary task probe stimulus.
2. Experiment
1
Experiment 1 investigated the effects of miscuing on dishabituation to and expectancy of S2 on the trial following miscuing (re-presentation trials). All subjects received a series of Sl-S2 pairings intermixed with S3-alone presentations. For the miscuing group (group E), there were two occasions on which S2 was miscued by its presentation following S3. Miscuing did not occur for the control group (group C). The S2 for both groups was an electric shock. Because previous research on the effects of miscuing has employed stimuli of only moderate intensity, level of shock was manipulated. For half the subjects in each of groups E and C, shock was “clearly discernible”, whereas for the other half it was “uncomfortable, but not painful”. Thus, the design was a 2 x 2 (miscuing x Shock level) factorial. 2. i. Method 2.1.1. Subjects The subjects were 48 undergraduate volunteered after providing informed
students (age range 18-50 years) who consent and who received course credit
for participation. Participants were restriction that all groups contained women.
allocated randomly to groups with the the same proportion (3 : 9) of men and
2.1.2. Apparatus Skin conductance was recorded directly by applying a constant voltage of 0.5 V across domed Ag-AgCI electrodes in conjunction with 0.05M NaCl electrolyte. It was recorded from masked areas on the distal phalanges of the index and second fingers of the subject’s non-preferred hand using a Grass 7PlG preamplifier and a chart speed of 2.5 mm/s. Recording sensitivity was 0.05 pS/mm of pen deflection. Respiration was recorded using chest pneumograph bellows (Phipps & Bird, Model RPBA) in conjunction with a Grass 7Pl G preamplifier. Sl and S3 were black geometric shapes (circle and triangle) on yellow and blue backgrounds respectively. They were projected through a small window in the wall of the experimental chamber using a Pradovit projector fitted with a tachistoscopic shutter (Gerbrands, Model G1166). Stimuli were projected onto a screen situated 180 cm in front of the subject at cyc level. Intensity was 534 cd/m’ as measured by an exposure photometer (Salford Electrical Instruments), and projected image size was 192 cm x 72 cm. The shape which served as Sl and S3 was counterbalanced within each group. Stimulus durations and intertrial intervals were controlled by an IBM-compatible microcomputer. Shock (S2) was produced by a locally constructed constant voltage stimulator (O-90 V peak) that delivered pulsed shocks. Pulse duration was 2 ms and the repetition rate was 50 Hz. Shock was delivered via a concentric electrode (diameter 36 mm) in which contact between the skin and metal elements was made through sponge pads soaked in normal saline (Tursky, Watson, & O’Connell, 1965). The electrode was attached to the volar surface of the left forearm, 17 cm from the ulna styroid. Prior to electrode application, the site was prepared by using EC2 electrode cream (Grass Instruments) to ensure stability of the electrode-skin interface. Expectancy of S2 was measured by means of a small dial and pointer positioned on the preferred arm of the chair in which subjects sat. The pointer could be rotated through 180” from “certain that shock will occur” through “uncertain” to “certain that shock will not occur”. The pointer was spring loaded so that it returned to the 90 o (uncertain) position when released. The position of the dial and pointer unit was adjustable both horizontally and vertically to ensure a comfortable operating position. Movement of the pointer was recorded on one channel of the polygraph with a range of - 20 mm to + 20 mm of pen deflection. 21.3. Procedure After attachment work-up procedure.
of the shock electrodes, They were asked to report
subjects underwent a shock when they felt the shock level
D.A. T Siddle et al. / Miscuing
und dishahituation
233
to be (a) clearly discernible and (b) uncomfortable, but not painful. Shocks of 1 s duration were then delivered in 0.5-V increments commencing at 0 V, with the subject reporting following the delivery of each shock. Although all subjects underwent the same work-up procedure, the level was then set at “clearly discernible” for half the subjects in groups E and C and at “uncomfortable, but not painful” for the other half. Following shock work-up, subjects were seated in a semi-reclining padded chair in the sound-attenuated experimental room. Ambient illumination was 2 lux, the temperature within the range 19-24°C and the relative humidity within the range 49-55%. The stimulus equipment and recording apparatus were housed in an adjoining control room. Subjects were informed that the first part of the experiment involved a 3-min rest period during which they were to relax with their eyes open. Following the pre-stimulation period, subjects were informed that they would see some shapes on the screen and that some shocks would occur from time to time. The operation of the expectancy unit was explained and subjects were asked to indicate their moment-to-moment expectation that shock was about to occur. Examples of different pointer positions were provided, and subjects were given the opportunity to ask questions. They were informed that the scale on the dial was continuous so that the pointer could be placed in any position and that they could move it as often as they liked or keep it in the one position for as long as they liked. Subjects in group C received 23 Sl-S2 pairings intermixed with 23 S3-alone presentations. Subjects in group E received 21 Sl-S2 pairings, two S3-S2 pairings (miscuing trials) and 21 S3-alone presentations. Miscuing occurred on trials 11 and 22. For both groups, Sl-S2 trials were presented at randomly ordered intervals of 40, 45 and 50 s (offset of S2 to onset of Sl). Sl and S3 were each 4 s in duration, and S2 duration was 1 s. Onset of S2 was coincident with the offset of Sl (or S3 on miscuing trials). S3-alone presentations occurred in the intervals between Sl-S2 pairings with the restriction that zero, one, and two S3 presentations occurred with equal probability per interval. An S3-alone presentation did not occur, however, in the intertrial intervals immediately following miscued trials. 2. I. 4. Scoring SCRs greater than 0.05 PS which occurred within l-4 s following Sl, S2, or S3 onset were scored as stimulus-evoked responses, and magnitude data were subjected to a square root transformation prior to analysis. Non-specific response frequency during the pre-stimulation period was quantified by counting the number of SCRs greater than 0.05 pS. Skin conductance level was calculated immediately prior to the presentation of Sl on the re-presentation trials. Expectancy of S2 during Sl and S3 was assessed on each trial by measuring pen position 500 ms prior to the onset of S2.
234
D.A. T. Siddle et al. / Mircuing
and dishahituation
2.2. Results A rejection region of p < 0.05 was used and all main effects and interactions which involved repeated measures were examined using multivariate tests (Rao R). Preliminary analyses revealed no significant group differences in terms of number of non-specific responses displayed during the pre-stimulation period, F (3,461 < 1. However, high-shock groups (M = 50.06) received more intense shocks, F (1,44) = 21.78, MS, = 429.97, than did low-shock groups (M = 22.13). 2.2.1. Electrodermal
activity
2.2.1.1. Responses during trials l-10. Electrodermal responses to Sl, S2, and S3 during the first 10 trials were arranged into 5 blocks of 2 trials and analysed using separate 2 X 2 X 5 (Group X Shock X Block) multivariate analyses of variance. Responding to Sl, S2, and S3 declined significantly across blocks, R (4,41) = 6.00, MS, = 0.03; R (4,41) = 10.35, MS, = 0.03; R (4,41) = 7.34, MS, = 0.02, respectively. Responding did not vary as a function of miscuing group for any of the stimuli, Fs (1,441 < 1. Surprisingly, responses to S2 in the high-shock condition were not significantly different from those in the low-shock condition, F (1,44) = 1.10, MS, = 0.24. None of the interactions approached significance. 2.2.1.2. Miscued trials. Responses to S2 on miscued trials for groups E and C in the high- and low-shock conditions are shown in fig. 1 (left panel). These data were analysed using a 2 x 2 X 2 (Group X Shock x Trial) ANOVA. Group E was more responsive than group C, F (1,44) = 20.58, MS, = 0.15, and SCRs were larger to high shock than to low shock, F (1,44)= 6.39. Responding was also greater on the first miscued trial than on the second, F (1,44) = 4.25, MS, = 0.03. There was also a Group x Shock interaction, F (1,44) = 3.65, and further analysis revealed that although group E was more responsive than group C in the high-shock condition, F (1,44) = 20.78, the difference in the low-shock condition only approached significance, F (1,441 = 3.45, p = 0.07. Thus, although miscuing produced enhanced SCR magnitude, the effect was more pronounced in the high-shock condition than in the low. 2.2.1.3. Re-presentation trials. Responses to S2 on re-presentation trials are shown in fig. 1 (right panel). The data were analysed using a 2 X 2 X 2 (Group x Shock x Trial) ANOVA. SCRs were larger in group E than in group C, F (1,44) = 10.83, MS, = 0.07 and larger in the high-shock condition than in the low, F (1,44) = 4.41. Responding was greater on the first re-presentation trial than on the second, F (1,44) = 8.83. MS, = 0.03, and
235
0.8
a
Group E
q
GroupC
0.6
0.4
0.2
Hi shock
Lo shock CONDITION
0.5 Group E
b EI]
GroupC
0.4
0.3
0.2
0.1
1st trial
2nd trial
TRlAL Fig. 1. Mean SCR magnitude to S2 as a function of group (miscuing vs. control) and shock intensity (high vs. low) on miscued trials (a) and as a function of group and trial on re-presentation trials 0~1.
there was also a Group X Trial interaction, F (1,44) = 9.82. Further analysis revealed that although group E was more responsive than group C on the first re-presentation trial, F (1,46) = 17.24, this was not the case on the second, F (1,46) = 1.48. Responses to Sl on re-presentation trials were analysed in a similar fashion to the S2 responses. Group E (M = 0.30) was more responsive F (1,44) = 7.16, MS, = 008, than group C (M = 0.14), and responding was greater, F (1,44) = 6.23, MS, = 0.04, on the first re-presentation trial (M = 0.27) than on the second (M = 0.17). 221.4. Skin cund~ct~nce ler,el. SCL immediateIy prior to Sl on the re-presentation trials was anaiysed using a 2 X 2 X 2 (Group X Shock X Trial) ANOVA. There were no significant main effects or interactions. 22.2. Expectancy data
2.2.2.1. Trials l-10. Expectancy of S2 in the presence of Sl and 53 was analysed using a 2 x 2 x 2 x 10 (Group x Shock x Stimuli X Trial) MANOVA. There were no effects for group or for shock condition (both Fs (1,44) < 1). However, expectancy of S2 was greater in the presence of Sl than in the presence of S3, F (1,44) = 346.64, MS, = 498.91. There was also a Stimuli x Trials interaction, R (9,36)= 26.72, MS, = 51.13, the nature of which is clear in fig. 2. The Stimuli X Shock interaction F (1,44) = 5.83, &KS, = 498, reflected the fact that differ~ntia1 expectancy was somewhat greater in the high-shock condition than in the low-shock. 2.2.2.2. Re-presentation trials. Expectancy of S2 in the presence of Sl on the two re-presentation trials was analysed using a 2 X 2 X 2 ANOVA. Expectancy was less in group E than in group C, F (1,44) = 3.82, MS, = 114, and less on the first re-presentation trial than on the second, F(1,44) = 5.82, MS, = 62.60. Analysis of the Group X Trial interaction, F(1,44) = 4.0, revealed that expectancy was lower in group E than in group C on the first re-presentation trial, F (1.46) = 5.38, but not on the second, F (1,44) < 1. There was also a Group x Shock interaction, F (1,44)= 6.38, and further analysis revealed that expectancy was less in group E than in group C in the high-shock condition, F (1,44)= 10.04, but not in the low-shock condition, F (1,44) < 1). The expectancy data from representation trials are shown in fig. 3 _. 2.3. Discussion The results of experiment 1 indicate that, consistent with previous data (Packer & Siddle, 1989; Siddle, 1985), SCR magnitude to a miscued stimulus
D.A.T. Siddle et al. / Miscuing and d&habituation
237
100 0
l
l
80 60 -
g z
40 -
b
20 -
M
SllHi
shock
v
Sl/Lo
shock
I
S3/Hi shock
P
S3/Lo shock
z O-
5 k;
-20 -
f
-40 -
z
-60 -80 _,oo
Fig. 2. Mean level.
I
I
I
I
I
1
I
I
I
I
1
2
3
4
5
6
7
8
9
10
S2 expectancy
TRIAL in the presence of Sl and S3 as a function
of trials and S2 shock
was greater than SCR magnitude in a control condition that did not involve miscuing. However, the effect was more marked when the miscued event was a high shock than when it was a low shock (fig. 1, left panel) and it was greater on the first miscued trial than on the second. In evaluating the effect of shock intensity, it should be noted that the effect of miscuing in the low-shock condition was not as marked as in previous work with stimuli of moderate intensity (Packer & Siddle, 1989). Although the appropriate psychophysical scaling has not been performed, it is possible that the low shock employed here was a subjectively weaker stimulus than the tones, lights, and vibration used previously. In any event, the most important effect - the dishabituation to S2 produced by the miscuing of S2 by S3 - was not affected by S2 (shock) intensity. However, the dishabituation effect was evident only on the first re-presentation trial (fig. 1, right panel). The expectancy data were consistent both with previous results and with the electrodermal data. Thus, expectancy of S2 in the presence of Sl increased across Sl-S2 pairings whereas expectancy in the presence of S3 decreased (see Packer et al., 1989). However, expectancy in the presence of Sl was disrupted by the miscuing of S2 by S3. Specifically, expectancy of S2
100 Group E f3
G:oupG
80 -
7
60 : .: ../. .:.: . ..
40 -
.:,
.
I
.:
“,
.. .. . . i ..:.
PO -
1.:.
.’
“:...
j .:,:
...
_.
:..: : ..,.,,
:;
i
Hi shock 1 st trial Fig. 3. Mean shock
S? expectancy
HI shock 2nd tnal on r~-pr~s~nt~lti~~n
Lo shock tst trial trials
Lo shock 2nd trial
for groups
E and C and for high and low SZ
level.
was lower in the miscuing condition than in the control on the first representation trial, but not on the second. Across both re-presentation trials, the difference in S2 expectancy between experimental and control conditions was greater in the high-shock condition than in the low. In summary, the results of experiment 1 indicate that the miscuing of S2 by S3 interfered with the associative link between Sl and S2. Moreover, the increased electrodermal responding to S2 on the trial following miscuing impIies that S2 was processed more elaborately in the miscuing groups than in the control on re-presentation trials. This possibility was examined further in experiment 2.
3. Experiment
2
The secondary task probe reaction time technique has been used in a number of studies in order to assess the extent to which stimuli which elicit orienting also command processing resources. In studies in which SlLS2 pairings have been presented, probe reaction time has been shown to be slowed when probes are presented during S2 omission, S2 re-presentation, and during a miscued S2 (Packer & Siddle, 1989: Siddle & Packer, 1987).
D.A.T. Siddle et al. / Miscuing and d&habituation
239
Packer and Siddle (1989) also reported that with Sl, S2, and S3 events of moderate intensity, reaction time was slowed to probes presented on the re-presentation trial following the miscuing of S2 by S3. However, it remains to be demonstrated that the same effect is obtained when a more intense S2 is employed. Experiment 2 examined this question. It utilized the same stimulus sequence as in experiment 1. Miscuing (group E) and control (group C) conditions were used, and the primary task involved Sl-S2 pairings intermixed with S3-alone presentations. Reaction time to a secondary task probe was the dependent variable. Because the effects of miscuing on dishabituation were most marked in the high-shock condition in experiment 1, only a high-shock condition was employed in experiment 2. 3.2. Method 3.2.1.
Subjects
The subjects were 36 undergraduate volunteers (age range 18-50 years) who participated after providing informed consent. Subjects were assigned randomly to one of two groups with the restriction that each group contained the same proportion (1: 2) of men and women. The data from one subject were excluded from analysis because of equipment malfunction. 3.2.2. Apparatus
The apparatus for stimulus generation and presentation was the same as in experiment 1. White noise of 70 dB was produced by a custom-built noise generator and was used as the secondary probe stimulus. The white noise was presented through stereophonic headphones (Telephonics TDH 49) and was calibrated (re: 20 pN/m2> by using a Bruel and Kjaer artificial ear 9 (type 4152) and a Bruel and Kjaer microamplifier (type 2603). Reaction time (ms) to probes was recorded by the computer as the time between probe onset and the activation of a microswitch held in the preferred hand. All stimulus durations and intertrial intervals were controlled by the microcomputer. 3.2.3. Procedure Following a shock work-up procedure and a 3-min pre-stimulation period, subjects were given standard instructions on the use of the microswitch. They were asked to press the switch as soon as possible after hearing the white noise. Practice trials consisting of 30 white noise probes were then delivered at randomly ordered intervals of 10, 20, and 30 s. Following the practice trials, subjects were informed that in the next part of the experiment they would see some shapes and receive a shock from time to time. They were asked to concentrate on the relationship between the shape and the onset of shock, but to operate the microswitch as quickly as possible whenever the noise occurred. The stimulus sequence was the same as in experiment 1, and
Table
I
Mean adjusted
response times in milliseconds
S2 on re-presentation
Trial
(and standard
crrora) for probes presented
during
trials
12
Trial 23
group E
group C
5 I9
421
(46)
(32)
39X
364
(41)
(31)
stimuli had the same parameter values. White noise probes occurred during 3 Sl events, 3 S2 events, and 3 S3 events. In addition, there were 6 probes during intertrial intervals, and in each case these occurred 5 s prior to the next Sl event. Probes also occurred during S2 on the two re-presentation trials (trials 12 and 23). Probe onset time was 300 ms after S2 onset and its duration was SO0 ms. Probes were randomly assigned to intertrial intervals and to stimuli with the restriction that probes occurred on trials 13-23 with the same probability as on trials 1- 12. 3.3 Results and discussion
Preliminary analysis of mean reaction time to probes presented during intertrial intervals revealed no difference between groups, F (1,341 < 1). Mean reaction time was 430 ms for group E and 462 ms for group C. In order to reduce intersubject variability, mean reaction time to intertrial interval probes for each subject was used as a covariate in subsequent analyses. Analysis of covariance of reaction time to probes presented during S2 on the representation trials (trials 12 and 23) in group E and to probes presented during S2 on corresponding trials in group C revealed a significant effect for group. F (1,33) = 4.12, MS, = 18616. There was also a significant effect for trials, F (1,33) = 9.09, MS, = 15715, but no Group X Trial interaction, F (1,33) < 1. Adjusted mean reaction time and standard error for each group and trial arc presented in table 1. The results of experiment 2 indicate clearly that reaction time to secondary task probes presented during S2 on re-presentation trials following miscuing is slower than in a control condition which does not involve miscuing. The data suggest, therefore, that S2 commands more processing resources on the trial following its miscuing.
4. General discussion The results of the present experiments are consistent with previous that a miscued event elicited enhanced electrodermal responding.
data in More
D.A. T. Siddle et al. / Miscuing and dishubituution
241
importantly, the miscuing of S2 by its presentation following a stimulus that had not previously predicted its occurrence produced dishabituation on the next trial when S2 again followed Sl. In addition, miscuing disrupted the extent to which S2 was expected in the presence of Sl. Finally, reaction time to secondary task probe stimuli presented during S2 on re-presentation trials was slower in the miscuing condition than in the control. To the extent that probe reaction time reflects the processing resources allocated to the primary task, the data indicate that S2 was processed more elaborately in the miscuing condition than in the control on re-presentation trials. Collectively, these results are similar to those reported by Packer and Siddle (1989) and indicate that the effects of miscuing are not confined to stimuli of moderate intensity, but can also be obtained with a more intense S2 of the sort used in studies of Pavlovian conditioning. Our expectancy data (fig. 2) indicate that the learning of the relationship between Sl and S2 was close to asymptote prior to the first miscuing trial, and on this basis it seems reasonable to conclude that the associative strength of Sl was near maximum for the S2 employed. On the trial following miscuing, however, expectancy of S2 in the presence of Sl was lower in the miscuing group than in the control. These data, together with the fact that miscuing produced dishabituation and that reaction time to a probe presented during S2 on the re-presentation trial was slower in the miscuing condition, suggest that miscuing disrupted the associative link between Sl and S2. As we have argued previously, this result seems problematical for theoretical accounts of associative learning in which associative strength is gained or lost through reinforcement and non-reinforcement, respectively (e.g., Rescorla & Wagner, 1972). Subjects in the present experiments did not receive Sl in the absence of S2 (i.e., an extinction trial), and any loss in Sl associative strength must have come about because of a gain in S3 associative strength. We know, for example, that the miscuing of S2 by S3 results in an increase in S3 associative strength as measured by expectancy of S2 (Packer et al., 19891. How an S3-S2 trial can interfere retrospectively with an already-developed Sl-S2 association is not clear from current theories of associative learning. According to these theories, the only thing that subjects will learn from a miscuing trial is that S2 can follow both Sl and S3. Our data suggest, however, that the miscuing of S2 by S3 results in a resetting of subjects’ expectancies such that Sl is no longer perceived as a predictor of s2. The pattern of results raises the question of whether a rule-based approach might provide a better account than can associative theories. Indeed, a comprehensive theory of conditioning which relies on a rule-based performance system has been proposed by Holyoak, Koh, and Nisbett (1989). Although rule-based approaches have rarely been pitted against associative accounts, some relevant data were reported by Shanks (198.5). In a series of
242
D.A. T. Siddle et 01. / Miscuing and dishahituatim
studies, Shanks examined the development of judgements about action-outcome contingencies. He concluded that an associative theory was better able to account for his results than was a rule-based analysis. However, the rules investigated by Shanks were considerably less sophistocated than the theoretical system proposed by Holyoak et al. (1989), and it is clear that a rule-based account of human associative learning warrants further examination. Some of the effects observed here were more marked with high shock than with low, and appear, therefore, to be consistent with an associative account. Differential expectancy of S2 in the presence of Sl and S3 was greater in the high-shock condition than in the low (fig. 2), a result that is consistent with the notion that the amount of associative strength supportable by a stimulus is a function of the intensity of that stimulus. Moreover, electrodermal responding was greater on miscued trials in the high-shock condition than in the low. The pattern of electrodermal results is consistent with what might be expected if it is assumed that the expectancy data reflect differential associative strengths between high and low shock. That is. if the contingency between S3 and no-S2 was better learned in the high-shock condition, presentation of S2 following S3 will be a more “surprising” or “unexpected” event that will, according to Wagner (19781, be processed more elaborately and will elicit larger responses. One anomalous finding concerns the increased electrodermal responding to Sl in the miscuing condition on re-presentation trials. This result has not been obtained in any of our previous experiments on the effects of either S2 omission or S2 miscuing. One obvious explanation is that the miscuing of S2 produced sensitization (Groves & Thompson, 1970) which augmented responding. On the other hand, miscuing and control conditions did not differ in terms of skin conductance level immediately prior to the re-presentation trials. Alternatively, Hall and Pearce (1983) have argued that orienting reflects stimulus associability which is determined by the extent to which an event predicts its consequences; poor predictive accuracy produces high associability. Thus, the increased responding to Sl might reflect increased associability. The difficulty here is that the Hall and Pearce theory provides no mechanism by which the miscuing of S2 by S3 can influence the predictive accuracy of Sl. Because the enhanced responding to Sl following S2 miscuing appears to be an isolated finding, explanations of the effect should perhaps await replication. In conclusion, two further points should be noted. First, the miscuing of S2 in our procedure occurred with a stimulus (S3) that had been presented repeatedly in the absence of S2 so that subjects clearly had learned, prior to miscuing, that S3 was a predictor of the absence of S2. Whether an S3-S2 trial has the same effects when S3 has not been presented repeatedly in the absence of S2 remains to be seen. Second, the effects of miscuing were confined to the first re-presentation trial (see also Packer & Siddle, 1989).
D.A. T. Siddle et al. / Miscuing alrd dishabituatiorr
243
Thus, it seems that after the first re-presentation trial subjects learn that an S3-S2 miscuing trial will be followed by a further Sl-S2 presentation. Consistent with this interpretation, Packer and Siddle (1989) reported that miscuing disrupted expectancy of S2 in the presence of Sl only on the first of four re-presentation trials. Thus, it appears that in addition to trial-by-trial expectations about event occurrence, subjects develop higher order expectancies about the pattern of experimental events.
References Dawson. ME., Schell, A.M., Beers, J.R., & Kelly, A. (19821. Allocation of cognitive processing capacity during human autonomic classical conditioning. .Iuurnal of Experimental P~ychoiogyc General, 111, 273-295. Groves, P.M., & Thompson, R.F. (1970). Habituation: A dual-process theory. Psychologicul Rel’iew, 70, 419-450. Hall, G., & Pearce. J.M. (1983). Changes in stimulus associability during conditioning: Implications for theories of acquisition. In M. Commons, R. Hermstein, 6r A.R. Wagner (Eds.), Qf~u~ritor~1.e artu~y~j.~of behal,ior (Vol. 3, pp. 221-2391. Cambridge, MA: Ballinger. Holyoak, K.J., Koh, K., & Nisbett, R.E. (19891. A theory of conditioning: Inductive learning within rule-based default hierarchies. Psychological Review, 96, 315-340. Packer, J.S., & Siddle. D.A.T. (1989). Stimulus miscuing, electrodermal activity, and the allocation of processing resources. Psychophysiology, 26, 192-200. Packer, J.S., Siddle, D.A.T., & Tipp, C. (1989). Restoration of stimulus associability, electroderma1 activity, and processing resource allocation. Biological Psychology. 28, 105-121. Rescorla, R.A., &Wagner. A.R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A.H. Black & W.F. Prokasy (Eds.), Classical conditioning II: Current theory and research (pp. 64-99). New York: Appleton-Century-Crofts. Shanks. D.R. (19851. Continuous monitoring of human contingency judgment across trials. Memory and Cognition, 13, 158-167. Siddle, D.A.T. 11985). Effects of stimulus omission and stimulus change on dishabituation of the skin conductance response. Journal of ~perim~iltal ~~cho~ob~: Learning, lemon, und Cognition, II, 206-216. Siddle. D.A.T., Booth, M.L., & Packer, J.S. (19871. Effects of stimulus preexposure on omission responding and omission-produced dishabituation of the human electrodermal response. Quarterly Journul of Experimental P.yychology, 398. 339-363. Siddle, D.A.T., & Packer, J.S. (1987). Stimulus omission and dishabituation of the electrodermal orienting response: The allocation of processing resources. P~ych~~phy.sjo~~~~, 24, ISl- 190. Tursky, B.. Watson, P.D.. & O’Connell, D.N. (196.5). A concentric shock electrode for pain stimulation. Psychophysiokqy, 1, 582-591. Wagner, A.R. (19781. Expectancies and the priming of STM. In S.H. Hulse, H. Fowler, & W.K. Honig (Eds.1, Cognitke processes in animal behal’ior (pp. 177-2091. Hillsdale, NJ: Erlbaum.
229 0301-11511/90/$03..50
STIMULUS MISCUING AND DISHABITUATION: ACTIVITY AND ~SOURCE FLOTATION * David A.T. SIDDLE PACKER
* *, Daphne
BROEKHUIZEN,
ELECTRODERMAL
and Jeanette
S.
School of Behu~~iouralSciences, Macquarie University, New South Wales 2109, Australia
The present research investigated the effects of miscuing a shock stimulus on dishabituation of the skin conductance response and on the allocation of processing resources. In both experiments, a control group received 21 Sl-S2 pairings intermixed with 23 S3-alone presentations. For the experimental group, S2 was miscued on trials 11 and 22 by its presentation following S3. In Experiment 1 (N= 481, S2 was a shock that was either “clearly discernible” or ‘.uncomfortabl~ but not painful”. The results indicated increased electrodermal responding when S2 was miscued by S3 and subsequent dishabituation when S2 again followed Sl on the next trial. Miscuing produced dishabituation with both a high- and low-shock S2. A continuous measure of S2 expectancy revealed that expectancy of S2 in the presence of Sl declined as a result of miscuing. Experiment 2 (N = 36) employed reaction time to a white noise probe stimulus as the dependent variable. The critical data came from probes presented 300 ms following S2 onset on the SI-S2 trial immediately following miscuing. They indicated that miscuing produced slowed reaction time to probes presented during S2 on the following SI-S2 trial. Thus, the miscuing of S2 by S3 appears to result in an increase in processing resources devoted to S2. even when S2 is presented in its usual position following Sl. The results are discussed in terms of current theories of associative learning. Keywords:
Classical conditioning. allocation
Habituation,
Skin conductance,
Associative
learning,
Resource
1. Introduction Some theories of associative learning attempt to account for both habituation and Pavlovian conditioning within the same explanatory framework. For example, Wagner (1978) has argued that one process which underIies re* This research was supported by a grant from the Australian Research Council. The third author was supported by a National Research Fellowship. Thanks are due to Len Glue for assistance with computer programming and to Marie Ayre and Brett Daniels for assistance with data analysis. ** Requests for reprints should be addressed to Dr. David Siddle, Department of Psychology, University of Queensland, Queensland 4072, Australia.
sponse habituation is associatively generated priming. Associatively generated priming occurs as a result of associations developed between events such as contextual cues and the habituation stimulus. Contextual stimuli can then act as retrieval cues to prime into short-term memory (STM) a representation of the habituation stimulus. Similarly, a conditioned stimulus (CS) is said to act as a retrieval cue which primes a representation of the unconditioned stimulus (US) into STM. The extent to which an event is primed in STM is said to determine the degree to which that event is processed. Unprimed events are processed more eIaborately and lead to larger responses than do primed events. A considerable amount of recent data is consistent with an associative account of habituation. Fur example, Siddle and his colleagues have‘ employed a procedure in which paired stimuli (Sl and S2) are presented. They have reported that 52 omission after a number of Sl-S2 pairings results in electrodermal responding at the time of omission and in an increase in responding to a subsequent presentation of S2 when it is represented following Sl (Packer, Siddle, & Tipp, 1989; Siddle, Booth, & Packer. 1987). It can be argued that presentation of Sl-S2 pairings leads to the priming of S2 in STM by Sl. Omission of S2 is thus an unexpected event which is more elaborately processed and which leads to larger electrodermal responses. An SI-alone presentation also weakens the associative strength between Sl and S2 so that the priming of S2 by Sl is Iess effective on the following Sl-S2 trial. Thus, the occurrence of S2 is less expected, and so that event is more elaborately processed and elicits larger responses. There are two types of data consistent with this interpretation. First, response time to secondary task probe stimuli presented during S2 omission and S2 re-presentation is slower than in a no-omission control condition (Siddle & Packer, 1987). To the extent that performance on a secondary task reflects the processing resources allocated to a concurrent primary task (Dawson, Schcll, Beers, & Kelly, 1982), these data indicate that S2 omission and S2 re-presentation each command processing resources. Second, a continuous measure of S2 expectancy indicates that SI-S2 pairings resuIt in a negatively accelerated growth in S2 expectancy in the presence of Sl. In addition, omission of S2 produces a decrease in expectancy of S2 in the presence of Sl (Siddle et al., 1987). Not all data, however, are consistent with an associative analysis. For example, the miscuing of S2 by another event has also been shown to produce dishabituation of the electrodermal response to 52. Siddle (1985) and Packer and Siddle (1989) presented subjects with Sl-S2 pairings intermixed with presentations of S3-alone. On some trials, Sl was replaced by S3 so that S2 was miscued on those trials. Not only did the miscued S2 result in large electrodermal responses, but an S2 presented following Sl on the next trial also elicited larger responses than in a control condition. Moreover,
D.A. T. Siddle et al. / Miscuing and dishabituation
231
reaction time to probes presented during S2 on the miscuing and re-presentation trials was siower than in a control condition. Finally, the expectancy measure indicated that that the miscuing of S2 by S3 (i.e., an S3-S2 presentation) resulted in a decrease in the extent to which S2 was anticipated in the presence of Sl. What these data seem to indicate is that the miscuing of S2 by S3 interfered retrospectively with the associative link between Sl and S2. It is not clear how these results can be accommodated within theories of associative learning in that gains and losses in associative strength are usually said to occur only as a result of reinforcement or non-reinforcement of a CS (e.g., Rescorla & Wagner, 1972). One difficulty, however, is that the stimuli employed in the miscuing studies were tones, lights, and vibrations of moderate intensity. Stimuli of higher intensity and of the type usually employed in studies of Pavlovian conditioning have not been used, and it is possible that the effect of miscuing on dishabituation will not be reproduced when a typical US is employed as S2. The present experiments investigated this possibility. Experiment 1 examined the effects of S2 intensity on responses to the miscued stimulus and on dishabituation of the skin conductance response (SCR). A continuous measure of S2 expectancy was also employed in order to replicate the Packer and Siddle (1989) results which indicated that the miscuing of S2 by S3 reduced expectancy of 52 in the presence of Sl. Experiment 2 examined whether re-presentation following miscuing of an 52 of higher intensity commands processing resources as assessed by reaction time to a secondary task probe stimulus.
2. Experiment
1
Experiment 1 investigated the effects of miscuing on dishabituation to and expectancy of S2 on the trial following miscuing (re-presentation trials). All subjects received a series of Sl-S2 pairings intermixed with S3-alone presentations. For the miscuing group (group E), there were two occasions on which S2 was miscued by its presentation following S3. Miscuing did not occur for the control group (group C). The S2 for both groups was an electric shock. Because previous research on the effects of miscuing has employed stimuli of only moderate intensity, level of shock was manipulated. For half the subjects in each of groups E and C, shock was “clearly discernible”, whereas for the other half it was “uncomfortable, but not painful”. Thus, the design was a 2 x 2 (miscuing x Shock level) factorial. 2. i. Method 2.1.1. Subjects The subjects were 48 undergraduate volunteered after providing informed
students (age range 18-50 years) who consent and who received course credit
for participation. Participants were restriction that all groups contained women.
allocated randomly to groups with the the same proportion (3 : 9) of men and
2.1.2. Apparatus Skin conductance was recorded directly by applying a constant voltage of 0.5 V across domed Ag-AgCI electrodes in conjunction with 0.05M NaCl electrolyte. It was recorded from masked areas on the distal phalanges of the index and second fingers of the subject’s non-preferred hand using a Grass 7PlG preamplifier and a chart speed of 2.5 mm/s. Recording sensitivity was 0.05 pS/mm of pen deflection. Respiration was recorded using chest pneumograph bellows (Phipps & Bird, Model RPBA) in conjunction with a Grass 7Pl G preamplifier. Sl and S3 were black geometric shapes (circle and triangle) on yellow and blue backgrounds respectively. They were projected through a small window in the wall of the experimental chamber using a Pradovit projector fitted with a tachistoscopic shutter (Gerbrands, Model G1166). Stimuli were projected onto a screen situated 180 cm in front of the subject at cyc level. Intensity was 534 cd/m’ as measured by an exposure photometer (Salford Electrical Instruments), and projected image size was 192 cm x 72 cm. The shape which served as Sl and S3 was counterbalanced within each group. Stimulus durations and intertrial intervals were controlled by an IBM-compatible microcomputer. Shock (S2) was produced by a locally constructed constant voltage stimulator (O-90 V peak) that delivered pulsed shocks. Pulse duration was 2 ms and the repetition rate was 50 Hz. Shock was delivered via a concentric electrode (diameter 36 mm) in which contact between the skin and metal elements was made through sponge pads soaked in normal saline (Tursky, Watson, & O’Connell, 1965). The electrode was attached to the volar surface of the left forearm, 17 cm from the ulna styroid. Prior to electrode application, the site was prepared by using EC2 electrode cream (Grass Instruments) to ensure stability of the electrode-skin interface. Expectancy of S2 was measured by means of a small dial and pointer positioned on the preferred arm of the chair in which subjects sat. The pointer could be rotated through 180” from “certain that shock will occur” through “uncertain” to “certain that shock will not occur”. The pointer was spring loaded so that it returned to the 90 o (uncertain) position when released. The position of the dial and pointer unit was adjustable both horizontally and vertically to ensure a comfortable operating position. Movement of the pointer was recorded on one channel of the polygraph with a range of - 20 mm to + 20 mm of pen deflection. 21.3. Procedure After attachment work-up procedure.
of the shock electrodes, They were asked to report
subjects underwent a shock when they felt the shock level
D.A. T Siddle et al. / Miscuing
und dishahituation
233
to be (a) clearly discernible and (b) uncomfortable, but not painful. Shocks of 1 s duration were then delivered in 0.5-V increments commencing at 0 V, with the subject reporting following the delivery of each shock. Although all subjects underwent the same work-up procedure, the level was then set at “clearly discernible” for half the subjects in groups E and C and at “uncomfortable, but not painful” for the other half. Following shock work-up, subjects were seated in a semi-reclining padded chair in the sound-attenuated experimental room. Ambient illumination was 2 lux, the temperature within the range 19-24°C and the relative humidity within the range 49-55%. The stimulus equipment and recording apparatus were housed in an adjoining control room. Subjects were informed that the first part of the experiment involved a 3-min rest period during which they were to relax with their eyes open. Following the pre-stimulation period, subjects were informed that they would see some shapes on the screen and that some shocks would occur from time to time. The operation of the expectancy unit was explained and subjects were asked to indicate their moment-to-moment expectation that shock was about to occur. Examples of different pointer positions were provided, and subjects were given the opportunity to ask questions. They were informed that the scale on the dial was continuous so that the pointer could be placed in any position and that they could move it as often as they liked or keep it in the one position for as long as they liked. Subjects in group C received 23 Sl-S2 pairings intermixed with 23 S3-alone presentations. Subjects in group E received 21 Sl-S2 pairings, two S3-S2 pairings (miscuing trials) and 21 S3-alone presentations. Miscuing occurred on trials 11 and 22. For both groups, Sl-S2 trials were presented at randomly ordered intervals of 40, 45 and 50 s (offset of S2 to onset of Sl). Sl and S3 were each 4 s in duration, and S2 duration was 1 s. Onset of S2 was coincident with the offset of Sl (or S3 on miscuing trials). S3-alone presentations occurred in the intervals between Sl-S2 pairings with the restriction that zero, one, and two S3 presentations occurred with equal probability per interval. An S3-alone presentation did not occur, however, in the intertrial intervals immediately following miscued trials. 2. I. 4. Scoring SCRs greater than 0.05 PS which occurred within l-4 s following Sl, S2, or S3 onset were scored as stimulus-evoked responses, and magnitude data were subjected to a square root transformation prior to analysis. Non-specific response frequency during the pre-stimulation period was quantified by counting the number of SCRs greater than 0.05 pS. Skin conductance level was calculated immediately prior to the presentation of Sl on the re-presentation trials. Expectancy of S2 during Sl and S3 was assessed on each trial by measuring pen position 500 ms prior to the onset of S2.
234
D.A. T. Siddle et al. / Mircuing
and dishahituation
2.2. Results A rejection region of p < 0.05 was used and all main effects and interactions which involved repeated measures were examined using multivariate tests (Rao R). Preliminary analyses revealed no significant group differences in terms of number of non-specific responses displayed during the pre-stimulation period, F (3,461 < 1. However, high-shock groups (M = 50.06) received more intense shocks, F (1,44) = 21.78, MS, = 429.97, than did low-shock groups (M = 22.13). 2.2.1. Electrodermal
activity
2.2.1.1. Responses during trials l-10. Electrodermal responses to Sl, S2, and S3 during the first 10 trials were arranged into 5 blocks of 2 trials and analysed using separate 2 X 2 X 5 (Group X Shock X Block) multivariate analyses of variance. Responding to Sl, S2, and S3 declined significantly across blocks, R (4,41) = 6.00, MS, = 0.03; R (4,41) = 10.35, MS, = 0.03; R (4,41) = 7.34, MS, = 0.02, respectively. Responding did not vary as a function of miscuing group for any of the stimuli, Fs (1,441 < 1. Surprisingly, responses to S2 in the high-shock condition were not significantly different from those in the low-shock condition, F (1,44) = 1.10, MS, = 0.24. None of the interactions approached significance. 2.2.1.2. Miscued trials. Responses to S2 on miscued trials for groups E and C in the high- and low-shock conditions are shown in fig. 1 (left panel). These data were analysed using a 2 x 2 X 2 (Group X Shock x Trial) ANOVA. Group E was more responsive than group C, F (1,44) = 20.58, MS, = 0.15, and SCRs were larger to high shock than to low shock, F (1,44)= 6.39. Responding was also greater on the first miscued trial than on the second, F (1,44) = 4.25, MS, = 0.03. There was also a Group x Shock interaction, F (1,44) = 3.65, and further analysis revealed that although group E was more responsive than group C in the high-shock condition, F (1,44) = 20.78, the difference in the low-shock condition only approached significance, F (1,441 = 3.45, p = 0.07. Thus, although miscuing produced enhanced SCR magnitude, the effect was more pronounced in the high-shock condition than in the low. 2.2.1.3. Re-presentation trials. Responses to S2 on re-presentation trials are shown in fig. 1 (right panel). The data were analysed using a 2 X 2 X 2 (Group x Shock x Trial) ANOVA. SCRs were larger in group E than in group C, F (1,44) = 10.83, MS, = 0.07 and larger in the high-shock condition than in the low, F (1,44) = 4.41. Responding was greater on the first re-presentation trial than on the second, F (1,44) = 8.83. MS, = 0.03, and
235
0.8
a
Group E
q
GroupC
0.6
0.4
0.2
Hi shock
Lo shock CONDITION
0.5 Group E
b EI]
GroupC
0.4
0.3
0.2
0.1
1st trial
2nd trial
TRlAL Fig. 1. Mean SCR magnitude to S2 as a function of group (miscuing vs. control) and shock intensity (high vs. low) on miscued trials (a) and as a function of group and trial on re-presentation trials 0~1.
there was also a Group X Trial interaction, F (1,44) = 9.82. Further analysis revealed that although group E was more responsive than group C on the first re-presentation trial, F (1,46) = 17.24, this was not the case on the second, F (1,46) = 1.48. Responses to Sl on re-presentation trials were analysed in a similar fashion to the S2 responses. Group E (M = 0.30) was more responsive F (1,44) = 7.16, MS, = 008, than group C (M = 0.14), and responding was greater, F (1,44) = 6.23, MS, = 0.04, on the first re-presentation trial (M = 0.27) than on the second (M = 0.17). 221.4. Skin cund~ct~nce ler,el. SCL immediateIy prior to Sl on the re-presentation trials was anaiysed using a 2 X 2 X 2 (Group X Shock X Trial) ANOVA. There were no significant main effects or interactions. 22.2. Expectancy data
2.2.2.1. Trials l-10. Expectancy of S2 in the presence of Sl and 53 was analysed using a 2 x 2 x 2 x 10 (Group x Shock x Stimuli X Trial) MANOVA. There were no effects for group or for shock condition (both Fs (1,44) < 1). However, expectancy of S2 was greater in the presence of Sl than in the presence of S3, F (1,44) = 346.64, MS, = 498.91. There was also a Stimuli x Trials interaction, R (9,36)= 26.72, MS, = 51.13, the nature of which is clear in fig. 2. The Stimuli X Shock interaction F (1,44) = 5.83, &KS, = 498, reflected the fact that differ~ntia1 expectancy was somewhat greater in the high-shock condition than in the low-shock. 2.2.2.2. Re-presentation trials. Expectancy of S2 in the presence of Sl on the two re-presentation trials was analysed using a 2 X 2 X 2 ANOVA. Expectancy was less in group E than in group C, F (1,44) = 3.82, MS, = 114, and less on the first re-presentation trial than on the second, F(1,44) = 5.82, MS, = 62.60. Analysis of the Group X Trial interaction, F(1,44) = 4.0, revealed that expectancy was lower in group E than in group C on the first re-presentation trial, F (1.46) = 5.38, but not on the second, F (1,44) < 1. There was also a Group x Shock interaction, F (1,44)= 6.38, and further analysis revealed that expectancy was less in group E than in group C in the high-shock condition, F (1,44)= 10.04, but not in the low-shock condition, F (1,44) < 1). The expectancy data from representation trials are shown in fig. 3 _. 2.3. Discussion The results of experiment 1 indicate that, consistent with previous data (Packer & Siddle, 1989; Siddle, 1985), SCR magnitude to a miscued stimulus
D.A.T. Siddle et al. / Miscuing and d&habituation
237
100 0
l
l
80 60 -
g z
40 -
b
20 -
M
SllHi
shock
v
Sl/Lo
shock
I
S3/Hi shock
P
S3/Lo shock
z O-
5 k;
-20 -
f
-40 -
z
-60 -80 _,oo
Fig. 2. Mean level.
I
I
I
I
I
1
I
I
I
I
1
2
3
4
5
6
7
8
9
10
S2 expectancy
TRIAL in the presence of Sl and S3 as a function
of trials and S2 shock
was greater than SCR magnitude in a control condition that did not involve miscuing. However, the effect was more marked when the miscued event was a high shock than when it was a low shock (fig. 1, left panel) and it was greater on the first miscued trial than on the second. In evaluating the effect of shock intensity, it should be noted that the effect of miscuing in the low-shock condition was not as marked as in previous work with stimuli of moderate intensity (Packer & Siddle, 1989). Although the appropriate psychophysical scaling has not been performed, it is possible that the low shock employed here was a subjectively weaker stimulus than the tones, lights, and vibration used previously. In any event, the most important effect - the dishabituation to S2 produced by the miscuing of S2 by S3 - was not affected by S2 (shock) intensity. However, the dishabituation effect was evident only on the first re-presentation trial (fig. 1, right panel). The expectancy data were consistent both with previous results and with the electrodermal data. Thus, expectancy of S2 in the presence of Sl increased across Sl-S2 pairings whereas expectancy in the presence of S3 decreased (see Packer et al., 1989). However, expectancy in the presence of Sl was disrupted by the miscuing of S2 by S3. Specifically, expectancy of S2
100 Group E f3
G:oupG
80 -
7
60 : .: ../. .:.: . ..
40 -
.:,
.
I
.:
“,
.. .. . . i ..:.
PO -
1.:.
.’
“:...
j .:,:
...
_.
:..: : ..,.,,
:;
i
Hi shock 1 st trial Fig. 3. Mean shock
S? expectancy
HI shock 2nd tnal on r~-pr~s~nt~lti~~n
Lo shock tst trial trials
Lo shock 2nd trial
for groups
E and C and for high and low SZ
level.
was lower in the miscuing condition than in the control on the first representation trial, but not on the second. Across both re-presentation trials, the difference in S2 expectancy between experimental and control conditions was greater in the high-shock condition than in the low. In summary, the results of experiment 1 indicate that the miscuing of S2 by S3 interfered with the associative link between Sl and S2. Moreover, the increased electrodermal responding to S2 on the trial following miscuing impIies that S2 was processed more elaborately in the miscuing groups than in the control on re-presentation trials. This possibility was examined further in experiment 2.
3. Experiment
2
The secondary task probe reaction time technique has been used in a number of studies in order to assess the extent to which stimuli which elicit orienting also command processing resources. In studies in which SlLS2 pairings have been presented, probe reaction time has been shown to be slowed when probes are presented during S2 omission, S2 re-presentation, and during a miscued S2 (Packer & Siddle, 1989: Siddle & Packer, 1987).
D.A.T. Siddle et al. / Miscuing and d&habituation
239
Packer and Siddle (1989) also reported that with Sl, S2, and S3 events of moderate intensity, reaction time was slowed to probes presented on the re-presentation trial following the miscuing of S2 by S3. However, it remains to be demonstrated that the same effect is obtained when a more intense S2 is employed. Experiment 2 examined this question. It utilized the same stimulus sequence as in experiment 1. Miscuing (group E) and control (group C) conditions were used, and the primary task involved Sl-S2 pairings intermixed with S3-alone presentations. Reaction time to a secondary task probe was the dependent variable. Because the effects of miscuing on dishabituation were most marked in the high-shock condition in experiment 1, only a high-shock condition was employed in experiment 2. 3.2. Method 3.2.1.
Subjects
The subjects were 36 undergraduate volunteers (age range 18-50 years) who participated after providing informed consent. Subjects were assigned randomly to one of two groups with the restriction that each group contained the same proportion (1: 2) of men and women. The data from one subject were excluded from analysis because of equipment malfunction. 3.2.2. Apparatus
The apparatus for stimulus generation and presentation was the same as in experiment 1. White noise of 70 dB was produced by a custom-built noise generator and was used as the secondary probe stimulus. The white noise was presented through stereophonic headphones (Telephonics TDH 49) and was calibrated (re: 20 pN/m2> by using a Bruel and Kjaer artificial ear 9 (type 4152) and a Bruel and Kjaer microamplifier (type 2603). Reaction time (ms) to probes was recorded by the computer as the time between probe onset and the activation of a microswitch held in the preferred hand. All stimulus durations and intertrial intervals were controlled by the microcomputer. 3.2.3. Procedure Following a shock work-up procedure and a 3-min pre-stimulation period, subjects were given standard instructions on the use of the microswitch. They were asked to press the switch as soon as possible after hearing the white noise. Practice trials consisting of 30 white noise probes were then delivered at randomly ordered intervals of 10, 20, and 30 s. Following the practice trials, subjects were informed that in the next part of the experiment they would see some shapes and receive a shock from time to time. They were asked to concentrate on the relationship between the shape and the onset of shock, but to operate the microswitch as quickly as possible whenever the noise occurred. The stimulus sequence was the same as in experiment 1, and
Table
I
Mean adjusted
response times in milliseconds
S2 on re-presentation
Trial
(and standard
crrora) for probes presented
during
trials
12
Trial 23
group E
group C
5 I9
421
(46)
(32)
39X
364
(41)
(31)
stimuli had the same parameter values. White noise probes occurred during 3 Sl events, 3 S2 events, and 3 S3 events. In addition, there were 6 probes during intertrial intervals, and in each case these occurred 5 s prior to the next Sl event. Probes also occurred during S2 on the two re-presentation trials (trials 12 and 23). Probe onset time was 300 ms after S2 onset and its duration was SO0 ms. Probes were randomly assigned to intertrial intervals and to stimuli with the restriction that probes occurred on trials 13-23 with the same probability as on trials 1- 12. 3.3 Results and discussion
Preliminary analysis of mean reaction time to probes presented during intertrial intervals revealed no difference between groups, F (1,341 < 1). Mean reaction time was 430 ms for group E and 462 ms for group C. In order to reduce intersubject variability, mean reaction time to intertrial interval probes for each subject was used as a covariate in subsequent analyses. Analysis of covariance of reaction time to probes presented during S2 on the representation trials (trials 12 and 23) in group E and to probes presented during S2 on corresponding trials in group C revealed a significant effect for group. F (1,33) = 4.12, MS, = 18616. There was also a significant effect for trials, F (1,33) = 9.09, MS, = 15715, but no Group X Trial interaction, F (1,33) < 1. Adjusted mean reaction time and standard error for each group and trial arc presented in table 1. The results of experiment 2 indicate clearly that reaction time to secondary task probes presented during S2 on re-presentation trials following miscuing is slower than in a control condition which does not involve miscuing. The data suggest, therefore, that S2 commands more processing resources on the trial following its miscuing.
4. General discussion The results of the present experiments are consistent with previous that a miscued event elicited enhanced electrodermal responding.
data in More
D.A. T. Siddle et al. / Miscuing and dishubituution
241
importantly, the miscuing of S2 by its presentation following a stimulus that had not previously predicted its occurrence produced dishabituation on the next trial when S2 again followed Sl. In addition, miscuing disrupted the extent to which S2 was expected in the presence of Sl. Finally, reaction time to secondary task probe stimuli presented during S2 on re-presentation trials was slower in the miscuing condition than in the control. To the extent that probe reaction time reflects the processing resources allocated to the primary task, the data indicate that S2 was processed more elaborately in the miscuing condition than in the control on re-presentation trials. Collectively, these results are similar to those reported by Packer and Siddle (1989) and indicate that the effects of miscuing are not confined to stimuli of moderate intensity, but can also be obtained with a more intense S2 of the sort used in studies of Pavlovian conditioning. Our expectancy data (fig. 2) indicate that the learning of the relationship between Sl and S2 was close to asymptote prior to the first miscuing trial, and on this basis it seems reasonable to conclude that the associative strength of Sl was near maximum for the S2 employed. On the trial following miscuing, however, expectancy of S2 in the presence of Sl was lower in the miscuing group than in the control. These data, together with the fact that miscuing produced dishabituation and that reaction time to a probe presented during S2 on the re-presentation trial was slower in the miscuing condition, suggest that miscuing disrupted the associative link between Sl and S2. As we have argued previously, this result seems problematical for theoretical accounts of associative learning in which associative strength is gained or lost through reinforcement and non-reinforcement, respectively (e.g., Rescorla & Wagner, 1972). Subjects in the present experiments did not receive Sl in the absence of S2 (i.e., an extinction trial), and any loss in Sl associative strength must have come about because of a gain in S3 associative strength. We know, for example, that the miscuing of S2 by S3 results in an increase in S3 associative strength as measured by expectancy of S2 (Packer et al., 19891. How an S3-S2 trial can interfere retrospectively with an already-developed Sl-S2 association is not clear from current theories of associative learning. According to these theories, the only thing that subjects will learn from a miscuing trial is that S2 can follow both Sl and S3. Our data suggest, however, that the miscuing of S2 by S3 results in a resetting of subjects’ expectancies such that Sl is no longer perceived as a predictor of s2. The pattern of results raises the question of whether a rule-based approach might provide a better account than can associative theories. Indeed, a comprehensive theory of conditioning which relies on a rule-based performance system has been proposed by Holyoak, Koh, and Nisbett (1989). Although rule-based approaches have rarely been pitted against associative accounts, some relevant data were reported by Shanks (198.5). In a series of
242
D.A. T. Siddle et 01. / Miscuing and dishahituatim
studies, Shanks examined the development of judgements about action-outcome contingencies. He concluded that an associative theory was better able to account for his results than was a rule-based analysis. However, the rules investigated by Shanks were considerably less sophistocated than the theoretical system proposed by Holyoak et al. (1989), and it is clear that a rule-based account of human associative learning warrants further examination. Some of the effects observed here were more marked with high shock than with low, and appear, therefore, to be consistent with an associative account. Differential expectancy of S2 in the presence of Sl and S3 was greater in the high-shock condition than in the low (fig. 2), a result that is consistent with the notion that the amount of associative strength supportable by a stimulus is a function of the intensity of that stimulus. Moreover, electrodermal responding was greater on miscued trials in the high-shock condition than in the low. The pattern of electrodermal results is consistent with what might be expected if it is assumed that the expectancy data reflect differential associative strengths between high and low shock. That is. if the contingency between S3 and no-S2 was better learned in the high-shock condition, presentation of S2 following S3 will be a more “surprising” or “unexpected” event that will, according to Wagner (19781, be processed more elaborately and will elicit larger responses. One anomalous finding concerns the increased electrodermal responding to Sl in the miscuing condition on re-presentation trials. This result has not been obtained in any of our previous experiments on the effects of either S2 omission or S2 miscuing. One obvious explanation is that the miscuing of S2 produced sensitization (Groves & Thompson, 1970) which augmented responding. On the other hand, miscuing and control conditions did not differ in terms of skin conductance level immediately prior to the re-presentation trials. Alternatively, Hall and Pearce (1983) have argued that orienting reflects stimulus associability which is determined by the extent to which an event predicts its consequences; poor predictive accuracy produces high associability. Thus, the increased responding to Sl might reflect increased associability. The difficulty here is that the Hall and Pearce theory provides no mechanism by which the miscuing of S2 by S3 can influence the predictive accuracy of Sl. Because the enhanced responding to Sl following S2 miscuing appears to be an isolated finding, explanations of the effect should perhaps await replication. In conclusion, two further points should be noted. First, the miscuing of S2 in our procedure occurred with a stimulus (S3) that had been presented repeatedly in the absence of S2 so that subjects clearly had learned, prior to miscuing, that S3 was a predictor of the absence of S2. Whether an S3-S2 trial has the same effects when S3 has not been presented repeatedly in the absence of S2 remains to be seen. Second, the effects of miscuing were confined to the first re-presentation trial (see also Packer & Siddle, 1989).
D.A. T. Siddle et al. / Miscuing alrd dishabituatiorr
243
Thus, it seems that after the first re-presentation trial subjects learn that an S3-S2 miscuing trial will be followed by a further Sl-S2 presentation. Consistent with this interpretation, Packer and Siddle (1989) reported that miscuing disrupted expectancy of S2 in the presence of Sl only on the first of four re-presentation trials. Thus, it appears that in addition to trial-by-trial expectations about event occurrence, subjects develop higher order expectancies about the pattern of experimental events.
References Dawson. ME., Schell, A.M., Beers, J.R., & Kelly, A. (19821. Allocation of cognitive processing capacity during human autonomic classical conditioning. .Iuurnal of Experimental P~ychoiogyc General, 111, 273-295. Groves, P.M., & Thompson, R.F. (1970). Habituation: A dual-process theory. Psychologicul Rel’iew, 70, 419-450. Hall, G., & Pearce. J.M. (1983). Changes in stimulus associability during conditioning: Implications for theories of acquisition. In M. Commons, R. Hermstein, 6r A.R. Wagner (Eds.), Qf~u~ritor~1.e artu~y~j.~of behal,ior (Vol. 3, pp. 221-2391. Cambridge, MA: Ballinger. Holyoak, K.J., Koh, K., & Nisbett, R.E. (19891. A theory of conditioning: Inductive learning within rule-based default hierarchies. Psychological Review, 96, 315-340. Packer, J.S., & Siddle. D.A.T. (1989). Stimulus miscuing, electrodermal activity, and the allocation of processing resources. Psychophysiology, 26, 192-200. Packer, J.S., Siddle, D.A.T., & Tipp, C. (1989). Restoration of stimulus associability, electroderma1 activity, and processing resource allocation. Biological Psychology. 28, 105-121. Rescorla, R.A., &Wagner. A.R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A.H. Black & W.F. Prokasy (Eds.), Classical conditioning II: Current theory and research (pp. 64-99). New York: Appleton-Century-Crofts. Shanks. D.R. (19851. Continuous monitoring of human contingency judgment across trials. Memory and Cognition, 13, 158-167. Siddle, D.A.T. 11985). Effects of stimulus omission and stimulus change on dishabituation of the skin conductance response. Journal of ~perim~iltal ~~cho~ob~: Learning, lemon, und Cognition, II, 206-216. Siddle. D.A.T., Booth, M.L., & Packer, J.S. (19871. Effects of stimulus preexposure on omission responding and omission-produced dishabituation of the human electrodermal response. Quarterly Journul of Experimental P.yychology, 398. 339-363. Siddle, D.A.T., & Packer, J.S. (1987). Stimulus omission and dishabituation of the electrodermal orienting response: The allocation of processing resources. P~ych~~phy.sjo~~~~, 24, ISl- 190. Tursky, B.. Watson, P.D.. & O’Connell, D.N. (196.5). A concentric shock electrode for pain stimulation. Psychophysiokqy, 1, 582-591. Wagner, A.R. (19781. Expectancies and the priming of STM. In S.H. Hulse, H. Fowler, & W.K. Honig (Eds.1, Cognitke processes in animal behal’ior (pp. 177-2091. Hillsdale, NJ: Erlbaum.