Eehar. Res. Thu. Vol. 27, No. 2. pp. 161-166, 1989 OW5-7967 89 53 00 + 0.00 Printed in Great Britain. All rights resewed Copyright c 1989 Pergamon Press plc

SENSORY PRECONDITIONING AND UCS INFLATION IN HUMAN ‘FEAR’ CONDITIONING

KATE WHITE and GRAHAM C. L. DAWY

The City University, London, England

(Received 27 June 1988)

Sununnry-This study describes a human electrodermal conditioning experiment involving processes of sensory preconditioning and UCS inflation. In stage I of the experiment Ss received six presentations of a CS+ paired with an innocuous 65 dB tone (UCS) and six presentations of an unpaired CS - In stage 2, Ss in the experimental group had the aversiveness of the UCS inflated as the intensity of the 65 dB tone was increased to I15 dB. In stage 3, Ss were given test presentations of CSt and CS-. A differential CR to CS+ was found only in stage 3 of the experiment and only in Ss who had experienced the UCS inflation procedure. These results suggest that (i) sensory preconditioning had occurred in stage I despite the failure to observe a differential CR in this stage, and (ii) the differential CR observed in stage 3 was mediated by an internal representation of the UCS whose aversiveness had been inflated in stage 2. As well as confirming that processes of sensory preconditioning and UCS inflation can be observed in human as well as animal Ss, these findings have important implications for contemporary conditioning models of clinical fears. In particular, they suggest that a contemporary conditioning model of acquired fears is not bound by the need to discover contiguous stimulus-trauma experiences in the histories of clinical phobias since, in humans at least, processes of stimulus association and UCS revaluation appear to be relatively independent.

INTRODUCTION

Contemporary models of human Pavlovian conditioning suggest that Pavlovian conditioning generates learned associations between the conditioned stimulus (CS) and the unconditioned stimulus (UCS). As a result of this association, CS presentation activates an internal representation of the UCS which in turn mediates the conditioned response (CR) (cf. Davey. 1983, 1987a, b, c, 1989). Such models of human conditioning are based on an integration of our knowledge of contemporary animal conditioning with the results of recent human conditioning studies (e.g. Davey, 1987c), and suggest that Pavlovian conditioning in humans is a process reflecting relatively complex cognitive information processing of stimuli and associations. One important implication of this kind of cognitive model of human Pavlovian conditioning is that the strength of the CR can be modulated not just by the strength of the association between CS and UCS, but also by the Ss evaluation of the UCS. For instance, if the S has some post-conditioning experience which results in them revaluing an aversive UCS more favourably, then this will be reflected in a weakened fear CR to the CS (Davey and McKenna, 1983). Such findings have been regularly reported in the animal conditioning literature (e.g. Rescorla, 1980; Dickinson, 1980) but have only recently been extrapolated to humans.

It is the process of UCS revaluation that differentiates contemporary models of human conditioning from more traditional contiguity-based associative models. That is, nonassociative processes involving the revaluation of the UCS appear to be more important in modulating CR strength than associative factors which link the CS and UCS (cf. Davey, 1987a, 1989). The processes which can give rise to UCS revaluation in human subjects are quite varied and range from simple experience with the UCS itself (e.g. habituation), verbal or socially transmitted information about the UCS (such as information about changes in its aversiveness or probability of occurrence), and self-observation of the strength of the responses elicited by the CS and UCS (cf. Davey, 1987a, 1989; Davey and McKenna, 1983; Cracknell and Davey, 1989). These UCS revaluation processes can not only lead to a devaluation of the UCS representation which results in a weakened CR, but also, under appropriate conditions, to inflation of the UCS representation which results in a greater magnitude CR.

Davey (1989) has outlined a number of implications that processes of UCS revaluation have for conditioning models of clinical fears, and has described how a number of clinical phenomena that

I61

162 KAE WHITE and GRAHAM C. L. DAVEY

had hitherto been difficult to explain in conditioning terms can, with the help of UCS revaluation processes, be integrated into contemporary human conditioning theory. One such problematic factor was the apparent absence of trauma in many clinical anamneses. That is, many phobics were unable to recall any trauma at the time of the first appearance of their fear to the phobic stimulus (e.g. Rachman, 1977; Marks, 1969). Davey (1989) explains that this is not inconsistent with a contemporary conditioning account. For instance, in contemporary conditioning models the acquisition of the CS-UCS association and modulation of the aversiveness of the UCS are relatively independent processes. Thus, it is quite possible for an individual to learn an association between

the CS and UCS when the UCS is innocuous or relatively unaversive (this is known as sensory pre-conditioning, Prewitt, 1967; Tait, Marquis, Williams and Suboski, 1969; Rizley and Rescorla, 1972), and then have the aversiveness of the UCS subsequently inflated by individual experience with the UCS alone. In this conditioning scenario there are three stages, (i) the acquisition of an association between a CS and an innocuous UCS (this is known as ‘behaviourally-silent’ learning because the CS evokes no observable CR, cf. Dickinson, 1980) (ii) postconditioning inflation of the value of the UCS (by, for example, experience with similar UCSs of greater intensity), and (iii) further encounters with the CS which now evokes a CR because of the revalued aversiveness of the UCS representation. Under the circumstances of such a scenario a differential CR would be acquired in the absence of any overt pairings of the CS with a traumatic UCS.

The present study is an attempt to discover whether a combination of sensory preconditioning and UCS inflation can produce a differential aversive electrodermal CR in the absence of any overt pairings of the CS with an aversive UCS. If this is possible, then it is not a primary assumption of conditioning models of acquired fears that such fears should have a history of explicit pairings of the phobic situation with trauma.

METHOD

The Ss were 20 undergraduate volunteers of both sexes whose ages ranged from 18 to 35 yr. They

were all naive as to the purpose of the experiment and the majority were Ss with no knowledge of psychology. These Ss were randomly allocated to two groups labelled Group E (experimental, N = 10) and Group C (control, N = 10).

Apparatus

Skin conductance responses (SCRs) were measured by bipolar placement of 9 mm diameter

silver/silver chloride (Ag/AgCl) GSR electrodes to the medial phalanx of the first and third fingers of the left hand. Before the electrodes were attached the Ss fingers were cleansed with methyl alcohol and neptic electrode gel applied to the fingers. Electrodes were secured in place with elastoplast, and these electrodes projected changes in skin-conductance level (SCL) onto a Washington 400 MD 2C polygraph via a constant voltage circuit.

The experiment was conducted using a 8 k PET 2001 microcomputer interfaced with solid-state logic programming equipment located on a bus-bar rack system. Outline figures used as stimuli were projected onto the screen of the microcomputer. All Ss wore headphones which masked the ambient noise in the experimental room and could also present a 0.5-set burst of a 1000 Hz tone, the intensity of which could be varied between 65 and 115 dB.

Procedure

On introduction to the experimental room Ss were given written instructions telling them that

the experiment was designed to measure sweat gland activity. They were seated in a comfortable chair facing the microcomputer’s visual display unit (VDU) on which the experimental instructions and outline figures were to be displayed. Ss were then fitted with the GSR electrodes and headphones, and were told to expect brief instructions displayed on the screen, Ss were also informed that periodically they would be presented with the instruction “Press any key to continue” and they were informed of which key to press when this occurred. All Ss were initially told to relax and watch the VDU screen throughout the experiment. An adaptation period of approx. 1 min followed while instruments were calibrated to the skin conductance levels (SCLs) of individual Ss.

Sensory preconditioning and UCS inflation 163

After this adaptation period Ss pressed a key as instructed and the experiment proper began. The experiment consisted of three stages followed by a debriefing period in which each S was asked four questions about the experiment.

Stage I (Sensory preconditioning) Ss from both groups received identical training in which six presentations of a triangle (CS+) were projected onto the VDU screen followed immediately by an innocuous 0.5-set 65 dB tone (nominally the UCS). Randomly interspersed amongst these pairings were six presentations of an outline figure of a kitchen tap (CS-). This latter stimulus was a control stimulus which was never associated with the UCS. Both CS+ and CS- were presented for 6 set with a 20-set intertrial interval. The VDU screen was blank during all intertrial intervals. Triangle and tap were counterbalanced as CS+ and CS- across Ss.

Stage 2 (UCS injktion). During this second stage Ss were given 12 successive presentations of the 0.5~set tone (UCS) alone. The VDU screen remained blank during this stage and each UCS presentation was separated by a 5-see interval. For Group C the volume of the tone remained constant at 65 dB throughout all 12 presentations. However, for Group E the volume of this tone was increased across successive presentations from 65 dB on the first presentation to 115 dB on the last presentation. A 0.5-set 115 dB tone is known from other studies to act as a UCS which is effective in establishing an aversive CR (e.g. Davey and McKenna, 1983). For Group E the intensity of the tone was increased in the following steps: 65, 69, 74, 84, 94. 100, 104, 108, 112, 115, 115, 115 dB. The UCS was inflated in a step-by-step fashion in an attempt to ensure that at the end of this stage Ss perceived the tone as the same one they had experienced in stage 1, except, of course, it was now more aversive. The effectiveness of this step-by-step procedure in ensuring the integrity of the tone UCS was assessed in the post-experimental questionnaire.

Stage 3 (resting). This stage was identical for both groups and consisted of three presentations of CS+ alone and three presentations of CS- alone. Presentations of CS+ and CS- were randomised across Ss and there were no presentations of the tone during this stage. Each stimulus presentation was separated by a 20-set intertrial interval.

Post-experimental questionnaire. All Ss were asked to complete a post-experimental question- naire designed to facilitate interpretation of the SCR data obtained in the first three stages of the experiment. This questionnaire was designed to obtain (i) Ss’ ratings of the aversiveness of the UCS in stage I and at the end of stage 2 (on a IO-point scale), (ii) Ss’ interpretations of what was happening in the experiment, (iii) the views of Ss in Group E about the integrity of the UCS between stages 1 and 2, i.e. whether Ss perceived the altered intensity of the tone in stage 2 as being a change in volume of the original tone rather than the presentation of a new one, and (iv) whether Ss were expecting presentation of the UCS in stage 3.

Response measurement

In the case of a CR, a response was considered to be an upward curve of the SCR within 3 set of the onset of the stimulus, and the magnitude of the response was calculated by measuring the distance between the trough and the apex of the curve; all response changes which did not result in an upward curve of the SCR were given zero value. In the case of a UCR (measured during stage 2) a response was considered to be an upward curve of the SCR also within 3 set of the onset of the tone UCS. The vast majority of trials with responses possessed only one response peak; only a small number of trials exhibited two response peaks and when this occurred only the first response was measured.

Finally, all SCR values were treated with a quasi-correction factor used to aid comparison of SCR values across Ss. The formula for this is 4 = SCR/SCR,,,, (Lykken, 1972) and SCR,,,, was taken from the S’s largest response during phase 1 of the experiment.

RESULTS

In order to analyse the development of any differential responding across blocks of trials in stage 1, CRs were analysed in two blocks of 3 trials. Nevertheless, there were no significant differences between responding to CS+ and CS- either in the first 3-trial block or in the last 3-trial block of stage 1 (all F’s < 1). (see Table 1) Neither was there any significant difference between the first

164 KATE WHIFF and GRAHA!.~ C. L. DAVEY

0 s-

0.2-

Table I. Stage I Mean A&SCR (_tSEM) ovw blocks of 3 trials for CS + and CS - in Groups

E and C 0. ‘I-

Blocks of rriais _-- ,-J 7-- ,

1st 3 TRiALS LAST 3 TRIALS

Group E cs+ 0.11 kO.02 0.10~0.0? cs- 0.12+_0.05 0.07~0.01 Fig, I, Mean Ad SCR (+ SE.M) to the tone UCS for the first

Group c cs + 0.08 & 0.04 0.09 f 0.03 3 trials and last 3 trials of stage Z for Group E (open circles) cs- 0.07 i: 0.03 0.07 * 0.03 and Group C (solid circles).

and last 3-trial blocks for either CS + or CS - (again, all F’s < I). This suggests that the innocuous 65 dB UCS was ineffective in establishing a differential CR to CS+.

Figure 1 shows the effect of UCS inflation on Group E in comparison with Group C. This exhibited a significant group x trial block interaction [F(1,36) = 7.87. P < O.Ol], in which Group E showed a significant increase in SCR to the tone UCS between the first three and last three trials of this stage [t(9) = 2.978, P < 0.011, while Group C showed a significant decrease in SCR over this same period [t(9) = 2.937, P < O.Ol].

Figure 2 illustrates mean CR magnitude between stages 1 and 3 of the experiment for both groups of Ss. While responding in stage .I showed no significant effect of either group or stimulus (see above), responding in stage 3 did exhibit a significant group x stimulus interaction [F( 1,36) = 16.234, P -=I O.OOl]. While there was no difference in SCR magnitude between CSt and CS- for Group C [t(9) = 1.09, P > 0.11, Group E did exhibit greater response strength to CS+ than CS- [t(9) = 4.919, P < O.Ol]. Furthermore, a within-group comparison of responding between the last three trials of stage 1 and the three trials of stage 3 shows a significant stage x stimulus interaction for Group E [F(3,9) = 16.253, P < O.OOl] in which SCR magnitude to CS+ increased significantly between stages 1 and 3 [t(9) = 2.831, P < 0.011 while there is no difference in the magnitude of the SCR to CS - (t < 1). A similar within-group analysis for Group C exhibited neither a significant effect of stage nor stage x stimulus interaction (ail F's -c 1).

The post-experimental questionnaire revealed that (i) 95% of al1 Ss could verbalise the association between CSi- and the 65 dB tone in stage 1; (ii) at the end of stage 1 the mean ratings of tone aversiveness for Groups E and C were 2.3 and 1.9 respectively. At the end of stage 2 these ratings were 9.0 and 1.7 respectively. This suggests that the increase in UCR magnitude between stages I and 2 in Group E was accompanied by a corresponding increase in the Ss’ cognitive rating of the aversiveness of the tone; (iii) all Ss in Group E reported that in stage 2 they believed it was

0.4

a 0,3 ::

8 4 0.2 j

CS’ CS- CS' CS’

STAGE 1 STAGE 3

Fig. 2. Mean A#SCR (+SEM) to CS+ and CS- over the last 3 trials of stage I and the 3 trials of stage 3. Open blocks represent Group E, solid blocks represent Group C.

Sensory preconditioning and UCS inflation 165

the original tone UCS getting louder rather than a different louder tone being presented; and (iv) all Ss in both groups reported that they expected to hear the tone again in stage 3. Of the Ss in Group E, 9 out of 10 expected it to follow CS+ and be unpleasant. None of the Ss in Group C expected the tone to be unpleasant if it was presented again during stage 3.

DISCUSSION

The results of the present experiment suggest that (i) pairing a CS-t with an innocuous 65 dB tone UCS initially fails to produce a differential CR to the CS+, (ii) successful inflation of the UCS in stage 2 is demonstrated by both physiological (SCR) and cognitive measures of the aversiveness of the UCS, (iii) this postconditioning inflation of the UCS produces a differential CR to CS+ on subsequent testing. and (iv) the differential CRs found in stage 3 imply, first, that sensory preconditioning between CS+ and WCS had occurred in stage 1 (even though any learning was behaviourafly silent at that stage), and, second, that the differential CR found in stage 3 was mediated by a representation of the UCS which had been inflated during stage 2.

Most importantly from a theoretical point of view, these results suggest that a differential aversive CR can be acquired even though the CS has never been paired with an explicitly aversive UCS and that, in human conditioning at least, processes of stimulus association and response modulation are relativefy independent (cf. Davey, 1983, 1987a, 1989). The fact that post- conditioning UCS inflation generated a differential CR when none existed before also provides further inferential evidence that human Pavlovian conditioning is mediated by CS-UCS associ- ations, and that UCS revaluation is an important factor in the modulation of CR strength. This evidence is also consistent with the results of UCS inflation studies with animals, which, in general, also suggest that conditioned responding is mediated by the evaluation of an internal UCS representation (e.g. Rescorla, 1974; Bouton, 1984; Hendersen, 1985).

Results such as those from the present study have important implications for contemporary conditioning models of acquired fears and phobias. First, they suggest that contiguous pairing of the fear-eliciting situation and trauma does not have to be a primary assumption of a conditioning model. Secondly, they impIy that the process of sensory preconditioning applies to human as well as animal conditioning and this process can resuIt in the acquisition of associations between events without that association being manifested as differential CRs. Thus, the associative component of a learnt fear could be acquired well in advance of any overt fear response. Thirdly, while the present study is a laboratory-based one, clear analogies can be drawn between the effects observed in this study and clinical fears which appear to lack any obvious association between the fear-inducing situation and trauma in the clinical anamneses. As an example. an individual may witness an unknown person die of a heart attack on a bus or train; on future occasions, riding on public transport may evoke memories of this incident but no anxiety (behaviouralIy-silent learning). Subsequently, however, that individual may be present when a close friend or relative dies of a heart attack, thus inflating the aversive properties of heart attacks. This may then give rise to acute anxiety when riding on public transport. In this particular scenario, public transport has never been directly associated with anxiety-eliciting trauma, but the public transport phobia results from a prior learned association between public transport and heart attacks, and subsequent independent inflation of heart attacks as aversive events. However, while it can be argued that the extrapolation of laboratory studies to clinical fears need to be independently substantiated, what the present data clearly suggest is that a conditioning model of such fears is no longer bound by the need to discover contiguous stimulus-trauma experiences in the histories of clinical phobias.

REFERENCES

Bouton 1. (1984) Differential control by context in the inflation and reinstatement paradigms. J. e.rp. Psycho/. Anim. Eehao. Proc. 10, 56-74.

Cracknell S. and Davey G. C. L. (1988) The effect of perceived unconditioned response strength on conditioned responding in humans. Med. Sci. Res. 16, 1699170.

Davey G. C. L. (1983) An associative view of human classical conditioning. In Animal Models of Hmxzn Beharior (Edited by Davey G. C. L.) Wiley, Chichester.

166 KATE WHITE and GRAHAM C. L. DAVEY

Davey G. C. L. (1987a) An integration of human and animal models of Pavlovian conditioning: associations, cognitions and attributions. In Cognitive Processef and P~[ocia~ Conditioning in Hwnans (Edited by Davey G. C. L.) Wiley, Chichester.

Davey G. C. L. (1987b) Integrating contemporary models of animal and human conditioning: implications for behaviour therapy. Paper presented to the 17th Annual Meeting of the European Association for Behaviour Therapy, Amsterdam.

Davey G. C. L. (1987~) Integrating human and animal theories of conditioning. J. Psychophysiol. 1, 105-108. Davey G. C. L. (1989) A contemporary model of human Pavlovian conditioning: the role of UCS revaluation and its

implications for the psychopathology of clinical fears. Manuscript under review. Davey G. C. L. and McKenna I. (1983) The effect of postconditioning revaluation of CSI and UCS following Pavlovian

second-order electrodermal conditioning in humans. Q. Jf exp. Psychof. 35B, 125-33. Dickinson A. (1980) Conremporary Animal Learning Theory, Cambridge University Press, Cambridge. Hendersen R. W. (1985) Fearful memories: the motivational significance of forgetting. In Afict, Conditioning and

Cognitions: Essays on the Dererminams of Behavior (Edited by Brush F. R. and Overmier J. B.). Lawrence Erlbaum, Hillsdale, N.J.

Lykken D. T. (1972) Range correction applied to heart rate and GSR data. Psychophysiology 9, 373-382. Marks 1. M. (1969) Fears and Phobias. Academic Press, New York. Prewitt E. P. (1967) Number of preconditioning trials in sensory preconditioning using CER training. J. camp. Physiol.

Psychof. 64, 360-362. Rachman S. (1977) The conditioning theory of fear-acquisition: a critical examination. Behav. Res. Ther. IS, 375-387. Rescorla R. A. (1974) Effect of inflation following conditioning. J. camp. Physiol. Psychol. 86, 101-106. Rescorla R. A. (1980) Pavlovian Second-order Conditioning: Studies in Associative Learning. Lawrence Erlbaum, Hillsdale,

N.J. Rizley R. C. and Rescorla R. A. (I 972) Associations in second-order conditioning and sensory preconditioning. /. camp.

Physiol. Psycho!. 81, t-l I. Tait R. W., Marquis H. A., Williams R., Weinsten L. and Suboski M. D. (1969) Extinction of sensory pr~onditioning

using CER training. J. camp. Physiol. Psychoi. 69, 1X%172.