Extinction of aversive eliciting functions as an analog of exposure to conditioned fear: Does it alter avoidance responding?

Full reference: Luciano, C., Valdivia-Salas, S., Ruiz, F. J., Rodríguez-Valverde, M., Barnes-Holmes, D., Dougher, M. J., Cabello, F., Sánchez, V., Barnes-Holmes, Y., & Gutierrez, O. (2013). Extinction of aversive eliciting functions as an analog of exposure to conditioned fear: Does it alter avoidance responding? Journal of Contextual Behavioral Science, 2(3-4), 120-134.

Study type: Analogue laboratory experiment with derived relations

Background and objectives

Research on the extinction of conditioned fear has demonstrated that the extinction of conditioned fear responses results in disruption of avoidance learning both directly trained and derived/indirectly established. However, it is unknown whether exposure to the conditioned event effectively leads to extinction of avoidance when it has been acquired through derived relations. Research on derived relational responding over the past decade has documented that a stimulus may acquire aversive eliciting functions (and avoidance functions) indirectly through its relation to another stimulus whose aversive function was directly conditioned.

The present study had two main objectives: first, to replicate through relational derivation the acquisition and transfer of both fear-eliciting responses and avoidance/approach responses using procedures already known in the literature; second, to critically examine whether extinction of fear respondents reduces the likelihood of subsequent derived avoidance responses, as exposure analog theory would suggest. The researchers predicted that conditional discrimination training followed by respondent extinction would produce a reduction in avoidance while demonstrating the clinical relevance of understanding changes in aversive functions in therapy.

Method

Participants

Eighteen undergraduate students (13 female; age range: 19-25 years) from different majors (introductory psychology, law, mathematics) at Universidad de Almería participated in the experiment. None had previous experience with the procedures employed. They were recruited through class announcements and received 10 euros for participation. Initially, all participants read an informed consent form indicating that they would receive mild electric shocks and were instructed that they could discontinue participation at any time without loss of compensation. All provided informed consent.

Experimental design

The study consisted of an intra-subject design with multiple phases. A between-conditions design was used wherein participants were exposed to identical procedures in two different experimental contexts, thereby facilitating examination of context sensitivity of extinction. In Experiment 1, a similarity relation between contexts (white vs. green) was trained to maintain extinction of avoidance responses in the green context when extinction occurred in the white context.

Materials and apparatus

The experiment was conducted in a laboratory consisting of two individual experimental cubicles with observation space and containment room. A two-way mirror was used for observing participants. All visual stimuli were presented on an HP nx9010 laptop screen (15 inch color). Skin conductance was measured using Ag/AgCl electrodes placed on the palm and palmar side of the distal third of the first and third fingers of each participant's non-dominant hand. A BIOPAC constant-voltage electro-dermal recording system (0.5 V) was used. Electric shocks were delivered via two disposable rounded electrodes placed on the inner surface of the participant's non-dominant forearm. A shock generator was set at 20 V for shock selection tests, producing 450 ms duration per shock.

Procedure

The entire experiment was conducted in one session lasting approximately 150-180 minutes. All procedures were reviewed and approved by the Ethics Committee for Research with Human Participants at Universidad de Almería. Participants were run individually.

Phase 1: Conditional discrimination training. Participants were presented with a computer task in which they had to select the correct symbol from two symbols presented. Six members of two equivalence classes were used (Class 1: A1-B1-C1-D1-E1-F1; Class 2: A2-B2-C2-D2-E2-F2). The procedure was identical to that described in previous research on equivalence relations, except that sequential relation training was used instead of two-way conditional discrimination training. Learning proceeded until participants emitted two consecutive correct responses for each relation. Following A-B training completion, subsequent training proceeded until participants responded correctly to all new relations (A-B and B-C). Mixed trial blocks containing A-B, B-C, C-D, D-E, E-F, and equivalence relations were presented in random order. A one-to-many conditional discrimination training procedure was used to establish two equivalence classes of four members each (Class 1: A1-B1-C1-D1; Class 2: A2-B2-C2-D2).

Phase 2: Respondent avoidance/approach training. During this phase, A1 and B1 served as CSS+ (followed by shock), while A2 and B2 served as CSS- (not followed by shock) in the respondent avoidance/approach training condition. During this phase, the circle was displayed as a white circle (4.25 cm diameter) located at the top left of the screen for 6 s, and was replaced by one of four B stimuli or comparison stimuli on the screen 2 s after circle onset.

Phase 3: Test of respondent and avoidance/approach function transfer. This phase was designed to determine whether transfer occurred to D and F stimuli. Participants were first presented with a 4-trial block of trials with D1 and D2 stimuli, using the same shock procedure as Phase 2. Participants were informed that if they failed to emit avoidance/approach responses as expected according to trained relations, then shock termination would be administered following D1, and otherwise, shock termination in the presence of D2.

Phase 4: Respondent extinction with A1 and B1. This phase began immediately following Phase 3. As in previous respondent conditioning trials of Phase 2, A or B stimuli appeared in the middle of the screen for 6 s in the white context. The application of extinction procedure only with A1 and B1 was intended to have a within-subject effect of changing whether participants would stop showing avoidance in the presence of Class 1 stimuli. The criterion used was that across three trials, the average difference in SCL between Class 1 and Class 2 members during the last three trials of each class was less than 0.05 µS, thereby meeting termination of extinction procedure. Following this, Phase 5 was initiated.

Phase 5: Critical Test. The equivalence test evaluated whether extinction of aversive eliciting functions with A1 and B1 would alter avoidance responding to other equivalent related members (D1 and F1). In this phase, procedures were identical to Phase 3, except no feedback was provided in any trials. The experimenter read equivalence test instructions to the participant.

Phase 6: Equivalence test. Mutual and combinatorial equivalence relations were evaluated using the same stimulus equivalence tests employed in Phase 3, with the difference that no feedback was provided in any trials. The experimenter read equivalence test instructions to the participant, the main difference being that participants were informed there would be no feedback from this point forward.

Data analysis

Individual and group analyses of the data were conducted to examine the effect of respondent extinction on avoidance. The percentage of participants meeting the specified criterion in each phase was calculated. Skin conductance data analysis was performed individually, determining for each trial in Phases 2-5 the percentage of participants meeting the specified extinction criterion. Skin conductance data analysis was conducted following statistical criteria established previously in studies on derived respondent relations. Non-parametric Wilcoxon signed-rank tests were used when normal distribution was violated, and paired t-tests were used when normality was found. Thresholds of 0.5 and 0.8 were assigned to define small, medium, and large effect sizes, respectively.

Results

In Experiment 1, 14 of 18 participants met the operational criterion for respondent conditioning during Phase 2 (differential presentation of A1 and B1, showing differentiated avoidance/approach responses according to acquisition of aversive properties of stimuli). Of these 14 participants, only 11 (78.6%) produced avoidance/approach responses during Phase 3 presentations (Function Transfer Test). Six of these 11 participants who showed function transfer in Phase 3 proceeded to Phase 4 with the condition of having produced both avoidance responses to D1 and F1 and approach responses to D2 and F2 during initial block trials of Phase 3. Nine of the 10 participants meeting criteria were conducted in Phase 5 (Critical Test). Of these 9, only 3 (33.3%) failed to show avoidance responses in the presence of A1 and B1 during the Critical Test. The other 6 still produced avoidance responses despite having received extinction.

Statistical analyses revealed significant differences between percentages of avoidance responses produced by the 9 participants in Class 1 (M= 66.67, SD= 50) during Phase 2 vs. Phase 5 (M= 27.78, SD= 27.9; t(8)= -3.83, p = 0.000, d= -1.603). A significant difference was also found between SCLs in Class 1 stimuli (M= 0.036, SD= 0.054) during Phase 2 vs. Phase 5 (M= 0.014, SD= 0.017; t(8)= -2.61, p= 0.031, d= 1.15). However, a significant difference was observed between percentages of avoidance responses produced during Phase 4 (when B1 stimulus was presented during extinction procedure) compared with Phase 3 (where extinction had not occurred).

In Experiment 2, 24 of 25 participants met the respondent conditional discrimination criterion. The criterion to determine whether respondent conditioning was acquired was that SCLs to A1 and B1 (CSS+) were greater than SCLs to A2 and B2 (CSS-) with more than 50% difference in SCL measured at response onset. Of these 24, 15 (62.5%) showed differentiated conditioning SCLs (larger SCLs to A1 and B1). Subsequently, all 15 participants were conducted through the transfer procedure. Eleven of the 15 participants (73.3%) produced avoidance responses to D2 and/or approach responses to F2 during the first stimulus presentations, while only four (26.7%) showed transfer to derived stimuli D1, E1, and F1.

Statistical analyses revealed that percentages of avoidance responses to the first stimulus presentation of Class 1 (M= 98.66, SD= 3.12) was significantly greater than for Class 2 stimulus (M= 3.33, SD= 8.33; Z= -4.58, p = 0.000, d= 14.43). These data demonstrated that the difference between avoidance responses on the first test trial was significantly greater for Class 1 stimulus (M= 0.038, SD= 0.067) compared with Class 2 stimulus (M= 0.034, SD= 0.042; t= 0.28, p = 0.28) during the last three trials per class. Nine of the 10 participants (90%) who had shown extinction of conditioned respondent SCLs to A1 and B1 in the Critical Test still produced avoidance responses despite having received extinction.

Discussion and conclusions

The main findings of the study were that extinction of respondent responses (SCLs) to A1 and B1 resulted in disruption of avoidance responses in both experiments. However, extinction of aversive eliciting functions was not completely effective for all participants in reducing avoidance. Approximately one third of participants in Experiment 1 and the majority of participants in Experiment 2 continued producing avoidance responses following respondent extinction.

These results replicate previous findings on the transfer of elicited fear functions and derived avoidance in the stimulus equivalence literature, demonstrating that extinction of conditioned fear responses does not necessarily eliminate subsequent avoidance when it has been acquired through derived relations. The authors suggest that this dissociation between respondent fear extinction and avoidance may reflect verbal or rule-governed processes that interfere with traditional exposure. Future research should examine whether additional manipulations (such as context similarity training in Experiment 2, which was designed to potentiate the dissociation between fear and avoidance) can improve extinction of respondent avoidance.

In conclusion, although extinction in the white context was effective in reducing conditioned fear responses, the results suggest that extinction may not be sufficient as an isolated clinical strategy for derived avoidance, particularly when avoidance has been potentiated through the availability of opportunities to avoid. The findings have important implications for understanding how extinction processes work in exposure therapy and suggest that additional procedures may be necessary to produce changes in avoidance when it has been derivatively acquired.

Significance and contribution

This study provides an important contribution to understanding how extinction processes of elicited aversive functions interact with derived avoidance through equivalence relations. The findings demonstrate that extinction of conditioned fear responses does not automatically reduce avoidance when it has been derivatively acquired, suggesting that procedures beyond traditional exposure may be necessary to produce changes in avoidance patterns based on relational learning. This analysis has implications for understanding the mechanisms of action of exposure therapy and for designing more effective interventions in behavioral therapy.