ACT. Extinction of Aversive Eliciting Functions as an. Luciano_et_al_in_press

7/28/2019 ACT. Extinction of Aversive Eliciting Functions as an. Luciano_et_al_in_press

1/60

Author's Accepted Manuscript

Extinction of aversive eliciting functions as ananalog of exposure to conditioned fear: Does

it alter avoidance responding?

Carmen Luciano, Sonsoles Valdivia-Salas,Francisco J. Ruiz, Miguel Rodrguez-Valverde,Dermot Barnes-Holmes, Michael J. Dougher,Francisco Cabello, Vanessa Snchez, YvonneBarnes-Holmes, Olga Gutierrez

PII: S2212-1447(13)00011-2DOI: http://dx.doi.org/10.1016/j.jcbs.2013.05.001Reference: JCBS18

To appear in: Journal of Contextual Behavioral Science

Received date: 17 January 2012Revised date: 10 April 2013Accepted date: 1 May 2013

Cite this article as: Carmen Luciano, Sonsoles Valdivia-Salas, Francisco J. Ruiz,Miguel Rodrguez-Valverde, Dermot Barnes-Holmes, Michael J. Dougher,Francisco Cabello, Vanessa Snchez, Yvonne Barnes-Holmes, Olga Gutierrez,Extinction of aversive eliciting functions as an analog of exposure toconditioned fear: Does it alter avoidance responding?, Journal of ContextualBehavioral Science, http://dx.doi.org/10.1016/j.jcbs.2013.05.001

This is a PDF file of an unedited manuscript that has been accepted forpublication. As a service to our customers we are providing this early version of

the manuscript. The manuscript will undergo copyediting, typesetting, andreview of the resulting galley proof before it is published in its final citable form.Please note that during the production process errors may be discovered whichcould affect the content, and all legal disclaimers that apply to the journalpertain.

www.elsevier.com/locate/jcbs


2/60

1

Extinction of aversive eliciting functions as an analog of exposure to

conditioned fear: Does it alter avoidance responding?

Carmen Luciano (1), Sonsoles Valdivia-Salas (1), Francisco J. Ruiz (1)

Miguel Rodrguez-Valverde (2), Dermot Barnes-Holmes (3), Michael J. Dougher (4),

Francisco Cabello (5), Vanessa Snchez (1), Yvonne Barnes-Holmes (3), & Olga

Gutierrez (6).

(1) Universidad de Almera

(2) Universidad de Jan

(3) National University of Ireland

(4) University of New Mexico

(5) Universidad de Murcia

(6) Universidad de Barcelona

Running head: Altering avoidance responding

Address corresponding to: Carmen Luciano, Ph.D., Ed. A. Facultad Psicologa,

Universidad Almera, 04120 Almera, Spain. Email: [email protected] Phone: 34-950-

015260


3/60

2

Running head: Altering avoidance responding

Extinction of aversive eliciting functions as an analog of exposure to

conditioned fear: Does it alter avoidance responding?

Abstract


4/60

3

Exposure techniques rely on the assumption that the extinction of the classically

conditioned response (i.e., fear) will result in the disruption of limiting forms of

avoidance behavior, both directly trained and derived/indirectly established. This report

presents translational research that attempts to test this assumption in laboratory

conditions in two experiments with human volunteers. The procedure in both

experiments included six phases: (1) conditional discrimination training for the

formation of two 6-member equivalence classes; (2) classical conditioning of elicited

responses to Class 1 (A1/B1) and Class 2 (A2/B2) members in the white context,

followed by conditioning of avoidance/approach responses to Class 1/Class 2 members,

respectively, in thegreen context; (3) test for the transfer of avoidance/approach

functions and of eliciting respondent functions to D1/F1 and D2/F2 in thegreen

context; (4) extinction of classically conditioned responses to A1/B1 in the white

context; (5) test of the effects of respondent extinction on avoidance responding to the

A, B, D, and F stimuli in thegreen context; and (6) test of derived symmetry and

equivalence relations. Results show that after successful respondent extinction in the

white context, only 33.3% participants stopped showing avoidance behavior in the

green context, and that respondent elicitation was reinstalled during the test (Phase 5).

In Phase 4 of Experiment 2, in addition to undergoing respondent extinction,

participants were instructed that the white and green contexts were similar. Results

show that after successful respondent extinction in the white context during Phase 4,

only 10% participants stopped showing avoidance behavior in the green context, and

that respondent elicitation was almost eliminated during the test (Phase 5). We discuss

these findings and their applied implications.


5/60

4

Key words: aversive conditioning, respondent extinction, equivalence relations,

avoidance, transfer and transformation of functions, skin conductance, human fear.

Extinction of aversive eliciting functions as an analog of exposure to conditioned fear:

Does it alter avoidance responding?

Traditionally, conditioning-based approaches to the explanation of anxiety

disorders have assumed that a history of direct aversive conditioning is necessary for the

acquisition of fear and avoidance responses (Barlow, 2002). Although clinical data

indicate that fears often emerge in the absence of any identifiable aversive conditioning

(e.g., Rachman, 1977, 1991), and recent research in derived relational responding (see

Dymond & Roche, 2009) and associative learning (Field, 2006) has identified ways in

which fear and avoidance can be learned indirectly, it could be said that the assumption

still holds in general terms, with some aversive conditioning experience needed at some

point in the genesis of anxiety.

Conditioning-based approaches have led to the design of exposure techniques,

widely used in behavior therapy for the treatment of anxiety disorders (e.g., Barlow,

2002; Deacon & Abramowitz, 2004; Marks, 1981). Exposure therapy is based on the

assumption that repeated exposure to the feared object or event (conditioned stimulus),

produces the extinction of the aversively conditioned responses (i.e., fear) and, hence,

the reduction of their behavioral outcome, namely avoidance (e.g., Craske &

Mystkowski, 2006; Mowrer, 1960). In fact, preventing avoidance is the ultimate goal of

exposure therapy, as this behavioral process is considered a critical factor in the

etiology and maintenance of anxiety disorders (e.g., Barlow, 2002; Forsyth, Eifert, &

Barrios, 2006; Hayes, 1976; Hayes, Wilson, Gifford, Follette, &Strosahl, 1996).


6/60

5

During the last 15 years, research on the extinction of conditioned fear has

focused on the conditions in which exposure treatments work (e.g., Hermans, Craske,

Mineka, & Lovibond, 2006), showing that the introduction of inhibitory CSs as safety-

signals (e.g., the presence of the therapist during exposure) or that the use of safety

behaviors such as avoidance have a deleterious effect on fear extinction (e.g., Lovibond,

Davis, & OFlaherty, 2000; Lovibond, Mithcell, Minard, Brady, & Menzies, 2009).

Also, it has been shown that the extinction of conditioned fear responses is context

sensitive; that is, that a change of context typically produces renewal of already

extinguished fear responses (Neumann & Longbottom, 2008; Vansteenwegen, Dirikx,

Hermans, Vervliet, &Eelen, 2006), turning fear extinction into a difficult target. Besides

the observed difficulties in obtaining fear extinction, somewhat surprisingly the central

assumption underlying exposure treatments remains untested in laboratory conditions.

That is, there is no laboratory evidence that the extinction of fear responses in the same

context in which they were conditioned will alter subsequent avoidance responding in a

context in which there is an actual opportunity to avoid.

Research on relational responding during the last decades is successfully

addressing some of the limitations of traditional conditioning approaches to the

acquisition of fear and avoidance (see Dymond & Roche, 2009; Forsyth et al., 2006).

Specifically, there is evidence that a stimulus may acquire eliciting and avoidance

functions indirectly by virtue of its relation to another stimulus whose eliciting and

avoidance functions were acquired by direct conditioning. For instance, Dougher,

Auguston, Markham, Greenway, and Wulfert (1994) demonstrated that after training

and testing for two four-member equivalence classes (A1-B1-C1-D1 and A2-B2-C2-

D2) and pairing B1 to electric shocks, most participants showed higher skin

conductance to C1 and D1 than to C2 and D2 (for similar results, see Rodrguez-


7/60

6

Valverde, Luciano, & Barnes-Holmes, 2009). Even more interestingly, Dougher,

Hamilton, Fink, and Harrington (2007) found higher skin conductance to a non-

conditioned stimulus (say C) than to an aversively conditioned stimulus (say B), by

virtue of the derived relation of comparison previously established between both stimuli

(B is less than C). A similar transfer of function effect has been observed with

respondent extinction. Dougher et al. (1994) exposed participants to aversive

conditioning by pairing several elements of the same equivalence class (B1, C1, and

D1) with shock. Then, only one of the elements underwent extinction (i.e., was

presented repeatedly without shock). As a result, the remaining members of the class

failed to elicit responses in a subsequent test.

Avoidance-evoking functions may also transfer across members of the same

relational network. By using similar procedures to those in Dougher et al. (1994),

Auguston and Dougher (1997) trained avoidance responding in the presence of an

aversive conditioned stimulus (B1 paired with shock) and then observed that other

members of the same equivalence class (C1 and D1) evoked avoidance responding

although they had never been directly paired with shock. Also, Roche, Kanter, Brown,

Dymond, and Fogarty (2008) showed that the extinction of avoidance responding in the

presence of one element of an equivalence class transferred to other elements of the

same class. In summary, data show that respondent elicitation and extinction, as well as

avoidance-evoking functions and operant extinction of avoidance, may transform

according to equivalence and non-equivalence relations. These results attest how

importantly verbal processes are involved in human conditioning (e.g., De Houwer,

2009; Lovibond, 2006). To date, however, no study has addressed the impact of

respondent extinction of conditioned fear on avoidance behavior, either directly trained

or acquired by relational means. Indeed, to our knowledge, no published study has


8/60

7

analyzed the transfer of both respondent elicitation and avoidance-evoking functions

simultaneously (in the same task and with the same participants). Published work has

focused on either one or the other.

The present study attempts to fill this gap, with two goals: first, to design an

experimental analogue of the acquisition and derived transfer of both respondent fear-

elicitation and avoidance-evoking functions by adapting well-known laboratory

procedures within the research area of derived relational responding; second, and most

importantly, to examine whether respondent fear extinction will reduce the likelihood of

subsequent avoidance responding (as an analogue of exposure techniques). Two

experiments were conducted with electric shocks as unconditioned aversive stimulation

during conditioning phases. In contrast to what was done in previous studies, we

measured both elicited (skin conductance) andoperant (avoidance and approach)

responses throughout the procedure. In Experiment 1, we tested whether respondent

extinction in the same context in which fear responses had been acquired would lead to

the alteration of avoidance behavior in a different context (in which avoidance had been

trained). Given the low impact of this procedure on avoidance responding, in

Experiment 2 we trained a relation of similarity between the context in which

respondent extinction occurred and the context in which avoidance responses were

available.

EXPERIMENT 1

Method

Participants

Eighteen undergraduates (13 females; age range = 19-25) attending different

courses (e.g., introductory psychology, maths, law) at Universidad de Almera


9/60

8

participated in the experiment. None of them had previous experience with the

procedures employed in this study. They were recruited through in-class and on-campus

flyer announcements, and each of them received 10 Euro for participation. At the

beginning, all participants read and signed a consent form informing them that they

would receive mild shocks and that they were free to discontinue participation at any

time without having to give up the 10 Euro they received in return. Upon completion of

the tasks, participants were fully debriefed.

Setting, Apparatus, and Stimuli

The setting, apparatus, and stimuli involved in this series of experiments were

almost identical to those in Rodrguez-Valverde et al. (2009); thus the following

description will mainly focus on their different features. The experiment was run in a

laboratory consisting of two adjacent rooms (an experimental cubicle and an

observation room) with a two-way mirror for participant observation. All visual stimuli

were presented on an HP nx9010 laptop computer (15 in color screen). Skin

conductance was measured and recorded according to the constant voltage technique

(0.5V) of exosomatic recording (see Dawson, Schell, & Filion, 1990) through a

computerized physiological recording system (BIOPAC Instruments) with non-

polarizable Ag/AgCl finger electrodes attached to the palmar side of the distal phalanx

of the first and third fingers of the participants nondominant hand. An isolated square-

wave stimulator (Laffayette 82415-IS) was used for the delivery of constant voltage

electric shocks (450 ms duration) through two disposable adhesive round electrodes

attached to the inner surface of the participants non-dominant arm (see Rodrguez-

Valverde et al., 2009, p. 88).

The visual stimuli were 18 black shapes, each framed in a square white

background (see Figure 1), presented on a general black background. The size of the


10/60

9

stimuli was 88 cm2. For ease of communication each stimulus was designated with an

alphanumerical label (e.g., A1, A2, A3). Participants never saw these labels.

Procedure

All procedures were reviewed and approved by the Ethics Committee for

Research with Human Participants at Universidad de Almeria. Upon arrival at the

laboratory and completion of the consent procedures, participants were escorted to the

experimental room for the administration of the computer tasks. Experiment 1 consisted

of six phases, all conducted in one session that lasted 150 to180 min approximately (see

Figure 2). Participants were run individually.

Phase 1: Conditional discrimination training. Participants were presented with a

card containing the following instructions:

During this task, a sample symbol will appear at the top of the computer screen

followed by three more symbols along the bottom. Your job is to select the

correct symbol at the bottom given the one at the top by using the keyboard:

press the Z key to select the symbol on the left, the V key to select the symbol in

the middle, and the M key to select the symbol on the right. When your selection

is correct, the word Correct will appear on the screen. When your selection is

incorrect, the word Incorrect will appear on the screen. Your job is to produce

as many correct selections as possible.

The experimenter then asked the participant to summarize what she would have

to do during the task. If the participant did not describe her task correctly, the

experimenter repeated the instructions and asked again. Once participants understood

the instructions, the experimenter left the room and the task commenced. An arbitrary

linear matching-to-sample (MTS) procedure was employed to establish two 6-member


11/60

10

equivalence classes (Class 1: A1-B1-C1-D1-E1-F1, and Class 2: A2-B2-C2-D2-E2-F2).

We utilized 6 members per class so as to have a sufficient number of related stimuli to

test the indirect or derived functions, whereas the linear conditional discrimination

procedure was used in order to rule-out second or higher-order associative conditioning

as possible explanations (e.g., Smyth, Barnes-Holmes, & Forsyth, 2006).

The trained relations for both classes were A-B, B-C, C-D, D-E, and E-F. On

any given trial, one sample stimulus (e.g., A1) appeared centred in the top third of the

computer screen. Two seconds later, three comparison stimuli (e.g., B1, B2, and B3)

appeared in the lower third of the screen with one in the middle, one on the left side and

the other one on the right side. A third set of six stimuli designated with number 3 (i.e.,

A3, B3, C3, etc.) was used as incorrect comparisons in the procedure, with no explicitly

trained relations amongst them. The location of the comparison stimuli varied randomly

across trials. Participants selection cleared the screen and produced the written

feedback Correct or Incorrect. The feedback remained on the screen for 2 s, and an

inter-trial interval (ITI) of 2 s preceded the next trial.

The training sequence proceeded as follows. Each new relation (starting with

A1-B1) was trained until the participant emitted two consecutive correct responses.

Training with the same relational pair in Class 2 (e.g., A2-B2) followed until two

consecutive correct responses were produced. Subsequently, both relational pairs (e.g.,

A1-B1 and A2-B2) were presented in random order in blocks of four trials (two per

relational pair), until completion of one block with 100% correct selections. This same

sequence was repeated with the remaining relational pairs from each class (i.e., B1-C1,

B2-C2, C1-D1, C2-D2, D1-E1, D2-E2, E1-F1, and E2-F2). Blocks of mixed trial-types

(with equal number of trials for Class 1 and Class 2 relations) were interspersed as

follows. After B-C training was completed, 4-trial blocks containing A-B and B-C


12/60

11

relations (one trial per relational pair) were presented until production of two

consecutive blocks with 100% correct responses. After completion of C-D training, 16-

trial blocks containing four A-B, four B-C, and eight C-D trials were presented until

production of one block with 100% correct selections. After completion of D-E and E-F

training, 4-trial blocks containing D-E and E-F relations (one per relational pair) were

presented until production of two consecutive blocks with 100% correct selections. This

was followed by the presentation of 6-trial blocks containing the C-D, D-E and E-F

relations (one trial per relational pair) until two blocks with 100% correct responses

were completed. Finally, blocks containing all the trained relations (i.e. A-B, B-C, C-D,

D-E, and E-F) in random order were presented until completion of three consecutive 10-

trial (one per relation) blocks with 100% correct selections.

Phase 2: Respondent and avoidance/approach conditioning with A and B

stimuli. During this phase, A1 and B1 served as CSs+ (i.e., were followed by shock),

and A2 and B2 served as CSs- (i.e., they were followed by points, exchangeable upon

experiment completion for university canteen vouchers). Skin conductance responses

(SCRs, measured in S) and avoidance responses served as dependent variables.

Respondent conditioning. This part of Phase 2 started with a shock setup

procedure in order to select the shock level that would be used as unconditional

stimulation. Participants were fitted with the SCR recording and shock delivery

electrodes (see Setting, Apparatus, and Stimuli). They were told that the purpose of this

stage was to select an uncomfortable but not painful shock level to be used during this

and subsequent phases of the experiment. The shock generator was set to 20V (levels

ranged from 10 to 100V), and a brief shock was administered. If the participants did not

rate the shock as uncomfortable, the shock voltage was gradually increased in 20V steps

until the participants reported that the dispensed shock was uncomfortable but not


13/60


14/60

13

conductance to stabilize. During the ITI the screen remained black. A total of two 4-trial

(1 per stimulus, in random order) blocks were presented.

Avoidance/Approach training. Once the previous 8 conditioning trials were

completed the experimenter entered the experimental cubicle and read aloud the

following instructions:

From now on, you will at times have the opportunity to avoid the shock by using

the keyboard. Likewise, if you want to keep accumulating points, you will have

to use the keyboard. Those opportunities will be signalled by a change in the

appearance of the white circle located at the top left of the screen. At times, it

will progressively turn green. When the circle is completely green, then you can

avoid the shock by pressing the Q key with your free hand, and you can keep

accumulating points by pressing the P key. These keys will only be operative

when the circle is completely green. If you press before the circle has completely

turned green, the computer will count that as an error and you will have to start

over. Note as well that the keys will be operative for a very limited time after the

circle is fully green. This is a long phase and you have a limited number of Q

presses available, so use them only when you are sure that the shock will be

delivered. Not all stimuli are followed by shock, so it is important to respond in

accordance with what you have learned in the previous phase and with what you

will be learning in this one. Once again, it is extremely important to sit as still

as possible because any movement will disrupt the measurement of

physiological responses. Also, remember that if you find yourself becoming

upset at any time and would like to end your participation, you can call out and

I will stop the procedure. Do you have any questions?


15/60

14

Once the experimenter ensured participants understanding of the instructions,

participants pressed the space bar and the computer screen remained blank for 2 min for

the stabilization of skin conductance level (SCL). Then, the avoidance/approach

training started.

A typical operant trial was as follows: An A or B stimulus was presented on the

screen for 8 s. During the first 4 s, the circle at the top left of the screen remained white,

and neither the avoidance nor the approach key was functioning (see Figure 3). During

the last 4 s (transition interval) the circle was progressively filled with green and both

keys remained inoperative. The transition interval was intended to signal the upcoming

availability of the operant response once the colour transition was complete. During the

8 s interval, changes in skin conductance level were measured. Once the circle was

completely green, the avoidance and approach keys were operative for 1 s. During this

time, participants could avoid shock in the presence of A1 and B1 by pressing the Q key

(i.e., the avoidance response). Shock was delivered if they pressed no key or pressed the

wrong one (P). Likewise, they could earn points in the presence of A2 and B2 by

pressing P (i.e., the approach response). If a participant failed to emit the approach

response, then the message Number of Points: 0 was displayed for 1.5 s. After that,

the screen went blank for a 25-35 s ITI and a new trial began.

Training started with one 16-trial block containing 4 respondent conditioning

trials (one per stimulus; A1, B1, A2, B2) intermixed with 12 operant (avoidance and

approach) conditioning trials (3 per stimulus) presented in random order. Respondent

conditioning trials were intermixed in order to avoid the extinction of conditioned

physiological responses likely acquired during the first part of this phase. A block was

considered correct when participants produced avoidance responses on all A1/B1 trials

(and not on any A2/B2 trial) as well as when they produced approach responses on all


16/60

15

but two (maximum) A2/B2 trials (and never on A1/B1 trials). We adopted a less strict

criterion for the approach response because our main target was the acquisition and

extinction of aversive functions. After one correct block, participants were presented

with a second identical 16-trial block. Participants who failed to meet criterion in the

first block of mixed trials went through an extra respondent conditioning block (4 trials)

before the second block of mixed trials. Participants who failed at this second block too,

were dropped from further participation.

Phase 3: Test for the transfer of respondent and avoidance/approach functions

to D and F stimuli. Phase 3 began immediately after Phase 2 and it only included

operant trials. These were identical to the ones presented in Phase 2, with the sole

exception that instead of A and B, D and F stimuli were used. Participants were first

presented with one 4-trial block with D1 and D2 (2 trials per stimulus in random order).

If they failed to emit the avoidance and/or approach responses (as expected according to

the purportedly established equivalence relations), then consequences were delivered as

with the A and B stimuli (i.e., shock was administered after D1 and the zero-point

written feedback was displayed after D2). A 4-trial block with F1 and F2 (2 trials per

stimulus in random order) followed. The criterion for the transfer of avoidance was that

participants produced avoidance responses on all F1 trials andapproach responses on

all F2 trials. Participants who met the transfer criterion proceeded to the next phase;

participants who did not, were dropped from further participation.

Phase 4: Respondent extinction with A1 and B1. This phase started immediately

after Phase 3. As with respondent conditioning trials in Phase 2, A or B stimuli

appeared in the middle of the screen for 6 s in the presence of the white circle. The

difference was that during this phase neither A1 nor B1 were followed by shock, while

A2 and B2 were still followed by points. The application of the extinction procedure


17/60

16

only with A1 and B1 was intended to have a within-subject control of changes in

responding in the next phase of the experiment (i.e., to see if participants would stop

showing avoidance by not pressing the Q key in the presence of Class 1 stimuli, but

would still press the P key to obtain points in the presence of Class 2 stimuli).

Participants were presented with two 12-trial blocks (3 trials per stimulus in random

order).The extinction criterion was established as follows: the average difference in skin

conductance between Class 1 and Class 2 members during the last three trials of each

class should be less than 0.05 S. However, it was not possible to assess whether

participants had achieved this criterion until the whole procedure had finished.

Accordingly, all participants in Phase 4 proceeded to Phase 5.

Phase 5: Critical Test. Test for avoidance/approach functions after respondent

extinction. The purpose of this phase was to examine the effect of respondent extinction

with A1 and B1 on avoidance responding to these and other arbitrarily related

(equivalent) stimuli (D1 and F1). This phase started immediately following Phase 4.

Trials had the same format as the operant trials in Phase 2 (avoidance/approach

training), with the difference that shock was never presented, regardless of the

participants responses. First, a 2-trial block with the B stimuli was presented (first B1

and then B2; this sequence was maintained across participants). If participants produced

avoidance responses (i.e., pressed the Q key) in the presence of B1they were

immediately re-exposed to respondent extinction (two 4-trial blocks with A and B

stimuli, with the same format as in Phase 4). Subsequently they were presented with a

new 2-trial block with B1 and B2 to test for the avoidance response functions of B1.

This sequence of re-exposure to respondent extinction was repeated up to three times if

participants kept avoiding in the presence of B1. After that, they were dropped from

further participation. On the contrary, if participants did not show avoidance in the


18/60

17

presence of B1, they continued to a 2-trial block with A stimuli (first A2 and then A1,

with this sequence maintained across participants). As with the B stimuli, avoidance (in

this case with A1) entailed re-exposure to respondent extinction up to a maximum of

three times, and non-avoidance granted access to the next block (one 4-trial block with

the sequence B2, A1, B1, A2).

Only participants who did not show avoidance with A1 nor with B1 proceeded

to the test with D, E, and F stimuli. Those participants who had passed the transfer of

avoidance/approach test in Phase 3 (i.e., who had pressed the avoidance key on the first

D1 trial) were presented with the following test sequence: D1, D2, F1, E1, F2, E2.

Those participants who had not passed the transfer test in Phase 3 (and thus received

shock after the first presentation of D1) were presented with the following test

sequence: F1, F2, E1, E2, D2, D1. As with A1 and B1, no shock was delivered.

Phase 6: Equivalence test. Mutual and combinatorial relations were tested using

the same trial format as in conditional discrimination training (Phase 1), with the

difference that no feedback was provided in any trial. The experimenter read aloud the

following instructions:

As in a previous phase, you will see one stimulus at the top of the screen and

then three stimuli at the bottom. Please look at the stimulus at the top and then

choose one stimulus from the bottom by clicking on it with the mouse. There is

always a correct answer, but this time the computer will not tell you whether

your choice is correct or not. Answer according to what you have learned in a

previous phase and try to accumulate as many correct responses as possible.

The twenty possible combinatorial relations per class (i.e. 10 transitive and 10

equivalence relations: A-C, C-A, A-D, D-A, A-E, E-A, A-F, F-A, B-D, D-B, B-E, E-B,

B-F, F-B, C-E, E-C, C-F, F-C, D-F, F-D) were first tested in a 40-trial block (one per


19/60

18

relation, in random order). The test finished upon production of a minimum of 34

correct responses. Otherwise, a test for symmetry relations (five relations per class) was

presented next in one 10-trial block (one trial per relation in random order). Upon

production of at least eight correct responses, a new block of 40 combinatorial test trials

was presented. Participants failing to achieve the criteria in either the symmetry or the

second combinatorial block were deemed as not passing the equivalence test. After that,

the experiment finished.

Skin conductance response quantification

Response amplitude was the parameter selected for quantification according to

the following criteria: the largest increase in SCL (measured in Siemens [S]) was

calculated for each trial in Phases 2 to 5. This variation was measured from the point of

response onset to the highest SCL value within the following time periods:

(1) For Phases 2, 3, and 5, the 4-s period during which the circle transitioned

from white to green. Pilot work showed that once the avoidance contingencies were in

place, autonomic activation mainly took place during this period, and not during the

first 4-s interval (white circle).

(2) For Phase 4, during the 6-s period in which A and B stimuli were displayed

on the screen (always in the presence of the white circle), only if the response onset

point started at least 0.5 s after visual stimulus onset.

Negative variations in SCL during these measurement intervals were quantified

as zero.

Data Analysis

Individual and group analyses of the data were conducted to examine the effect

of respondent extinction on avoidance responding, both directly trained and acquired by

derived means. The primary datum for individual analyses was the percentage of


20/60

19

participants that met the specified criterion in each phase. Our main focus was on

individual adjustment to the successive manipulations performed through the

experiment, respondent extinction being the most important one. This reduced the

number of participants whose performance was suitable for analysis through the last

test. Specifically, during Phase 4 (respondent extinction), we only considered data from

participants who (1) had learned during Phase 2 (respondent and avoidance/approach

conditioning) to produce avoidance responses to elements of Class 1 and approach

responses to elements of Class 2 andalso had shown higher elicited SCRs to elements

of Class 1 than to elements of Class 2; and (2) during Phase 3 (Transfer Test) had

shown transfer of the avoidance/approach functions to Class 1/Class 2 members,

respectively. Likewise, in Phase 5 (Critical Test) we only analysed the performance of

participants for whom respondent extinction had been effective during Phase 4, in order

to examine the impact of successful respondent extinction on subsequent avoidance

responding. A detailed description of the specific achievement criteria and of the

number of participants whose data were analysed in each phase is presented in the

Results and Discussion section.

As mentioned previously in the procedure section, participants who did not meet

the operant criterion were dropped from further participation. It is important to note,

however, that the SCR data were available to the experimenters only upon completion

of the experimental tasks. The analysis of SCR data showed that some participants who

had met the operant criterion and thus completed all phases, had not met the respondent

conditioning or extinction criteria. Accordingly, their performance was not included in

the analyses. All the available individual data are presented in Appendix 1.

As to the analysis of group data, we calculated both the average number of

avoidance responses and the average SCRs to the elements of Class 1 and Class 2 in


21/60

20

each phase, for phases 2 to 5. All participants who produced avoidance and approach

responses correctly as expected according to the experimental design across phases (i.e.,

phases 2, 3, and 5) entered the analysis, regardless of whether or not they also met the

respondent conditioning or extinction criteria. This was intended to tracking the

synchronicity between elicited arousal responses and avoidance responding. One-

sample Kolmogorov-Smirnov tests were conducted on all pertinent SCR and avoidance

response datasets to determine if the data fulfilled the normality assumption. Where the

normality assumption was met, related samples T tests were conducted on avoidance

responses and on SCRs in order to establish comparisons between classes within the

same phase, and within classes across phases. Otherwise, the non-parametric Wilcoxon

Signed-Rank Test was conducted for this purpose. In each case, Cohens d was

calculated for statistically significant results in order to determine the effect size of

successive manipulations. Following Cohens (1988) guidelines, .2, .5, and .8 were used

as thresholds to define small, medium and large effects, respectively.

Results and Discussion

Conditional discrimination training(Phase 1)

All 18 participants met the training criterion. The total number of trials to

criterion varied across participants from 117 trials (P12) to 301 trials (P4) (see

Appendix 1, Phase 1).

Respondent and Avoidance/Approach Conditioning with A1 and B1 Stimuli (Phase 2)

The criterion to determine if respondent conditioning was acquired for each

participant was: larger SCRs to A1 and B1 (CSs+) than to A2 and B2 (CSs-) in more

than half of the conditioning trials, with an average difference of at least 0.05 S

between Class 1 and Class 2 stimuli. This criterion (adapted from Rodrguez-Valverde

et al., 2009) had to be achieved in any of the avoidance conditioning blocks, each of


22/60

21

which had 12 trials (six avoidance and six approach trials). As previously mentioned,

the operant conditioning criterion was that participants consistently pressed the

avoidance key in the presence of A1/B1 and the approach key in the presence of A2/B2.

Seventeen out of 18 participants (see Figure 4, Phase 2, and Appendix 1) met the

avoidance/approach criterion (94.4%). Of these, 14 (82.3%) met the respondent

conditioning criterion. Figure 5 (upper and lower graphs) shows that the percentage of

avoidance responses to Class 1 stimuli (M= 97.22, SD = 2.78) was significantly larger

than to Class 2 stimuli (M= 2.77, SD = 7.00;Z= -3.83,p< .000, d= 16.03); and that the

average SCR to Class 1 stimuli (M= .61, SD = .59) was significantly larger than to

Class 2 stimuli (M= .12, SD = .14; t= 4.13,p= .001, d= 1.13) for those participants

who met the avoidance/approach criterion (N= 17).

Transfer of Respondent and Avoidance/Approach functions to D and F(Phase 3).

The criterion to determine the occurrence of transfer relied entirely on the first

exposure to each test stimulus. The rationale for adopting this criterion has been

presented elsewhere (see Rodrguez-Valverde et al., 2009, pp. 96-97).

Transfer of Avoidance/Approach functions.Of the 14 participants who showed

both avoidance/approach and respondent conditioning during Phase 2, 11 (78.6%)

produced avoidance/approach responses during F1 and F2 presentations. Six of them

(54.5%) also showed transfer with D1 and D2: P17, P1, P10, P12, P5, and P16 (see

Figure 4, phase 3, and Appendix 1).

Statistical analyses (Figure 5, upper graph) showed that the percentage of

avoidance responding (pressing Q) to the first presentation of a Class 1 stimulus (M=

78.57, SD = 32.21) was significantly larger than to the first presentation of a Class 2

stimulus (M= 10.71, SD = 28.95;Z= -3.15,p = .002, d= 2.21).


23/60

22

Transfer of respondent elicitation functions. The criterion to determine that

transfer had occurred was similar to that established for respondent conditioning: larger

SCRs to D1 and F1 than to D2 and F2, respectively, during the first presentation of each

stimulus (adapted from Rodriguez-Valverde et al., 2009). Six participants out of the 11

(54.5%) who showed transfer of avoidance/approach functions also showed transfer of

respondent elicitation functions.

Statistical analyses revealed that SCRs to Class 1 stimuli (M= .90, SD = .93)

were larger (marginally significant) than to Class 2 stimuli (M= .36, SD = .39; t= 2.16,

p = .056, d= .76) for those participants who met the transfer of avoidance/approach

criterion (N= 11).

Respondent Extinction with A1 and B1 (Phase 4)

The criterion for respondent extinction was that the difference in average SCRs

between Class 1 and Class 2 stimuli was less than 0.05 S during the last three trials per

class. Nine out of the 10 participants who showed transfer of avoidance/approach

functions during Phase 3 (data from P12 were lost during the extinction phase) met the

respondent extinction criterion (see Appendix 1, Phase 4).

Statistical analyses showed that SCRs to Class 1 stimuli (M= .039, SD = .056)

did not differ from those to Class 2 stimuli (M= .036, SD = .054; t= -.15,p = .89)

during the last three trials (see Figure 5, lower graph). No participant emitted

avoidance/approach responses during extinction trials.

Critical Test: Effect of Respondent Extinction with A1 and B1 on Avoidance/Approach

(Phase 5)

Only three (P4, P5, and P6) out of the nine (33.3%) participants who had shown

extinction of elicited SCRs to A1 and B1did not show avoidance responding in the

presence of these stimuli during the critical test. The other six still produced avoidance


24/60

23

responses when presented with A1 and B1 (and four of them showed larger SCRs to

these stimuli too) in Phase 5. Approach responding to A2 and B2 remained intact for all

participants (see Figure 4 and Appendix 1, phase 5. Detailed data from all participants

are in Appendix 1).

Besides, in what regards the stimuli whose avoidance-evoking functions had

been indirectly acquired (i.e., D1, E1, and F1), only the data from two of the three

participants were available (P6 quit before the end of the experiment). Neither P4 nor

P5 showed avoidance responding in the presence of D1, F1, and E1.

Overall, statistical analyses revealed significant differences between the

percentages of avoidance responses produced by the nine participants to Class 1 (M=

66.67, SD = 50) and Class 2 stimuli (M= 0, SD = 0;Z= -2.5,p = .014, d= 1.89, see

Figure 5). SCRs during the first test trial were significantly larger to Class 1 stimuli (M

= .72, SD = .72) than to Class 2 stimuli (M= .12, SD = .17; t= 2.61,p = .031, d= 1.15).

A significant difference was also found between SCRs to Class 1 stimuli during

respondent extinction (Phase 4, last three trials in the presence of the white circle) and

SCRs to Class 1 stimuli during the Critical Test (Phase 5, first stimulus presentation in

the presence of the green circle) for all nine participants (N= 9, t= -2.89,p = .02, d=

1.34). The corresponding within-class comparison with Class 2 stimuli yielded no

significant changes across phases (t= -1.15,p = .28).

It is important to note that the Critical Test was programmed so as to drop

participants who continued avoiding in the presence of B1. As a result, we missed a

substantial amount of data regarding avoidance/approach responses and elicited SCRs

to the derived stimuli D, E, and F; the same applies to the equivalence class formation

data collected in Phase 6. Only the data for P4 and P5, who had survived to this phase,


25/60

24

were collected. In both cases, the equivalence criterion was met within the first block of

trials.

These results indicate that respondent extinction of conditioned SCRs to A1 and

B1 in the white context had very little effect on avoidance responding in the green

context. The analyses also show that more than half of the participants who ran through

the Critical Test showed fear renewal (measured as larger SCRs to Class 1 stimuli)

when moved from the white to the green context. Approach responses and small SCRs

to Class 2 stimuli, however, remained constant across contexts. This is consistent with

previous studies showing that respondent extinction is contextually controlled and that

conditioned fear is susceptible to the renewal effect when a change in context occurs

(Hermans et al., 2006; Vansteenwegen et al., 2006). It is likely that in the present

experiment, for most participants, the functions acquired by the white context during

respondent conditioning and extinction were different or opposite to the functions

acquired by the green context during avoidance conditioning. This might explain why

respondent extinction of SCRs in the white context did not generalize to the green

context for many participants, and why avoidance was still in place for most

participants. The participants informal post-task reports seem to support this

conclusion. For instance, some of the participants who continued to show avoidance

responding in the critical test said that they had pressed the key because that was what

they had done before in the green context. Others said that even though they were pretty

sure that the shock would not be delivered, they were pressing just in case.

With the same goal of Experiment 1, in Experiment 2 we trained a relation of

similarity between the white (i.e., respondent conditioning and extinction) context and

the green (i.e., avoidance) context, in order to examine whether this might facilitate

preventing avoidance responding upon a history of respondent extinction.


26/60

25

EXPERIMENT 2

As mentioned above, it is likely that most participants in Experiment 1 produced

avoidance responses during the Critical Test because the extinction and the avoidance

contexts had different functions, which probably resulted in a fear renewal effect.

Experiment 2 was designed to prevent such difference between contexts by training a

relation of similarity between the white and the green circle (i.e., the extinction and the

avoidance context, respectively). Extrapolating to the applied arena, this might be

analogue to instructing an anxious client that the context where they are undergoing

exposure therapy and the context where they usually show pathological forms of

avoidance are similar, so that similar reactions would be expected in both cases.

Additionally, in Experiment 2 all participants who got to the Critical Test went

on to complete the experiment, regardless of their performance in the first trials of the

Critical Test (i.e., whether or not they had avoided in the presence of B1 or A1).

Method

Participants

Twenty-five undergraduates (13 females; age range = 18-26) were recruited,

welcomed, compensated for participation, and debriefed as in Experiment 1.

Procedure

The procedures employed in Experiment 2 were almost identical to those in

Experiment 1, with two exceptions: (a) a procedure for assessing and training a relation

of similarity between the white and the green contexts was introduced in Phase 4; and

(b) unlike in Experiment 1, all participants who got to the Critical Test (Phase 5)

completed it, including test trials with the directly conditioned stimuli (i.e., A and B)

and with the derived stimuli (i.e., D, E, and F),and also completed the test for the

formation of equivalence classes.


27/60

26

Phase 4. Assessment and training of a similarity relation between the green and

white contexts. After the test for the transfer of respondent and avoidance/approach

functions with D and F Stimuli (Phase 3), there was an assessment and subsequent

training of the similarity relation between the green and the white contexts. The

assessment included both open-ended and close-ended questions. First, there were four

open-ended questions: the computer screen showed A1 or B1 or A2 or B2 in the

presence of either the green or the white circle, with the stimuli arranged as they

appeared during conditioning trials (Experiment 1, Phases 2 and 3). The experimenter

then asked: What came to mind immediately upon seeing this? and wrote down the

participants responses. The close-ended format assessment was next. The experimenter

asked participants to pay attention to the screen and to select the most appropriate

response option by clicking on it with the mouse, and then left the experimental room.

Figure 6 shows the typical arrangement of stimuli on the computer screen during any

given close-ended assessment trial. At the top third of the screen, a message reminded

participants to select the most appropriate response option according to their experience

in previous phases. Below, two pictures were displayed in the middle of the screen. One

of them depicted either A1 or B1 in the presence of the white circle, and the other one

depicted the same stimulus in the presence of the green circle (see Figure 6). The

response options (they are similar, they are opposite, or neither) were located

below, along the lower third of the screen (with their position varying randomly across

trials). Participants selections cleared the screen for 2 s, and were followed by the start

of a new trial. The close-ended assessment consisted of two trials, one per stimulus (i.e.,

A1 and B1). After that, participants underwent training of a similarity relation between

the green and the white circles. The trial format was the same as during close-ended

assessment, with two exceptions. First, an initial message instructed participants to keep


28/60

27

selecting one of the three options and to accumulate as many consecutive correct

responses as possible. Second, correct selections (i.e., they are similar) cleared the

screen and produced the written feedback Correct for 2 s; incorrect selections

produced the feedback Wrong. Training proceeded in blocks of two trials (one per

stimulus, A1 and B1, in random order) until participants produced two consecutive

blocks with 100% correct responses.

Once the similarity training was over, participants underwent respondent

extinction with A1 and B1 as in Experiment 1. In this case, however, upon termination

of the first 12-trial extinction block, and before the second 12-trial extinction block,

similarity blocks (two trials per stimulus: A1 and B1) were inserted in order to retrain

the similarity relation between the green and white contexts. They were repeatedly

presented until the participant responded correctly on all trials across two consecutive

blocks.

Phase 5: Critical Test. Effect of Respondent Extinction with A1 and B1 (plus

context similarity training) on Avoidance/Approach. This test was conducted as in

Experiment 1, with the only difference that both the elicited SCRs and

avoidance/approach responses to theD, F, and E stimuli were tested regardless of

whether participants had previously shown avoidance responses to the directly

conditioned A1 and B1 in this same test (see Experiment 1, Phase 5, for sequence

details).

Results and Discussion

Conditional Discrimination Training

All 25 participants met the training criterion. The number of trials necessary to

reach the criterion varied from 89 (P12) to 300 trials (P16) (see Appendix 2, Phase 1).

Respondent and Avoidance/Approach Conditioning with A1 and B1 stimuli


29/60

28

As shown in Figure 7 (see also Appendix 2, Phase 2), 24 participants (96%) met

the avoidance/approach criterion. That is, they avoided both A1 and B1, and approached

both A2 and B2 (see criterion in the Results section of Experiment 1). Of these 24, 15

(62.5%) showed differential SCR conditioning (larger SCRs to A1 and B1 than to A2

and B2).

Figure 8 (upper and lower graphs) shows that the percentage of avoidance

responses to Class 1 stimuli(M= 98.66, SD = 3.12) was significantly larger than that to

Class 2 stimuli (M= 3.33, SD = 8.83; Z = -4.58, p < .000, d = 14.43); also, that SCRs to

Class 1 stimuli (M= .11, SD = .13) were significantly larger than those to Class 2

stimuli (M= .04, SD = .05;Z= -3.94,p< .000, d= .71) for participants who met the

avoidance/approach criterion (N= 24).

Transfer of Respondent and Avoidance/Approach Functions to D and F

As in Experiment 1, only the results of the participants who met both the

avoidance/approach and respondent criteria during the conditioning phase (15 out of 24)

were analysed; however, the results for the nine remaining participants who achieved

the operant criterion are detailed in Appendix 2.

Transfer of avoidance/approach functions. Eleven out if 15 participants (73.3%)

produced avoidance responses to F1 and/or D1, and approach responses to F2 and/or D2

(P3, P5, P6, P16, P17, P21, P22, and P25, with both the F and D stimuli) (see Figure 7

and Appendix 2).

Statistical analyses (Figure 8, upper graph) showed that the percentage of

avoidance responding to the first presentation of Class 1 stimuli (D1 and F1;M= 85.71,

SD = 30.56) was significantly larger than toClass 2 stimuli (D2 and F2;M= 3.57, SD =

13.36;Z= -3.36,p = .001, d= 3.48).


30/60

29

Transfer of respondent elicitation functions. A shown in Figure 7, five of the 11

participants (45.8%) who had shown transfer of avoidance functions also showed larger

SCRs to the first presentation of D1 and F1 relative to the first presentation of D2 and

F2. However, statistical analysis yielded no significant differences between SCRs to

Class 1 stimuli (M= .15, SD = .10) and to Class 2 stimuli (M= .11, SD = .13; t=.92,p

= .38). That is, transfer of avoidance and approach responses to non-directly

conditioned Class 1 and Class 2 stimuli occurred even in the absence of a differential

pattern of elicited arousal responses between classes.

As in Experiment 1, the data for all participants who met the operant criteria

(avoidance/approach responding) are presented in Appendix 2.

Assessment and Training of Similarity between Green and White Contexts, and

Respondent Extinction with A1 and B1

The results obtained in the assessment of the relation between the white and

green contexts showed great variability across participants. After training the similarity

relation between contexts, however, all 11 participants selected the option they are

similar (100% correct, see criterion). As for respondent extinction with A1 and B1, all

but one of the 11 participants (90.9%) who had shown transfer of avoidance/approach

functions met the respondent extinction criterion (see Appendix 2, Phase 4). That is, the

difference between average SCR to Class 1 and Class 2 stimuli was less than .05 S

during the last three trials per class.

Statistical analyses showed that SCRs to Class 1 stimuli (M= .038, SD = .067)

did not differ from those to Class 2 stimuli (M= .034, SD = .042; t= .28) during the last

three trials for these 10 participants (see Figure 8).

Critical Test: Effect of Respondent Extinction with A1 and B1 on Avoidance/Approach

and Respondent functions


31/60

30

As shown in Figure 7 (see also Appendix 2, phase 5), only one (P8) out of the 10

participants (10%) who had shown extinction of conditioned SCRs to A1 and B1

stopped showing avoidance responding in the presence of the same stimuli during the

Critical Test. The remaining 9 participants still showed avoidance responding when

presented with A1 and B1. However, only one participant (P25) showed larger SCRs to

A1 and B1 than to A2 and B2. As in Experiment 1, the approach response to the stimuli

in Class 2 remained intact for all participants.

As for the test for avoidance/approach responding to the D, F, and E stimuli,

participants showed an identical pattern of responding to that observed with A and B

(see Figure 7). The available data (one participant quit before the end, and data from

another two participants were lost; see Appendix 2 for details) show that the only

participant who had not shown avoidance responding with A1 and B1 (P8), also did not

with D1, F1, and E1; the six participants who showed avoidance responding in the

presence of A1 and B1, also did in the presence of D1, F1, and E1. All participants

produced approach responses in the presence of D2, F2 and E2, and none of them

showed larger SCRs to Class 1 stimuli than to Class 2 stimuli.

Statistical analyses revealed that the percentage of avoidance responses to A1

and B1 (N= 10; Class 1:M= 90, SD = 31.62) was significantly larger than that to A2

and B2 (M= 0, SD = 0;Z= -3,p = .003, d= 4.03). Likewise, the percentage of

avoidance responses to F1, D1, and E1 (N= 7: Class 1:M= 85.71%, SD = 37.79) was

significantly larger than that to F2, D2, and E2 (M= 0, SD = 0;Z= -2.5,p = .014, d=

3.24) (see Figure 8, upper and lower graph). Also, SCRs to A1 and B1 (N= 10;M=

.033, SD = .047) did not differ from those to A2 and B2(M= .018, SD = .014; t= .81,p

= .45); likewise, SCRs to D1, F1, and E1 (N= 7;M= .027, SD = .019) did not differ

from those to D2, F2, and E2 (M= .042, SD = .045; t= -1.31,p = .24). Lastly, no


32/60

31

significant differences between phases (Phase 4, respondent extinction: last three trials

in the presence of the white circle, vs. Phase 5, Critical Test: first stimulus presentation

in the presence of the green circle) were found for SCRs to Class 1 stimuli (t= -1.09,p

= .318). In other words, there was no renewal of elicited SCRs when changing from the

extinction context (i.e., white circle) to the avoidance context (i.e., green circle).

Although data from participants who did not show respondent conditioning in

Phase 2 were not considered in the analyses, it is interesting to note that six out of the

eight who were exposed to the Critical Test (see Appendix 2) produced avoidance

responses to both A1 and B1, and to the derived stimuli D1, E1, and F1 (although they

still showed no evidence of respondent conditioning in this test).

While a comparison of these data with those obtained in Experiment 1 is feasible

only for the first stimulus presentation per class in the Critical Test (note that

participants who produced avoidance responses to the first B1 presentation during the

Critical Test in Experiment 1 were not tested any further), we may conclude that

training the similarity relation between the white and the green context served to

maintain extinction of SCRs in the green context, from the first to the last trial, for all

but one of the 10 participants who made it to the Critical Test (see Appendix 2). That is,

the similarity instruction prevented fear renewal when moving from the white to the

green context. Most importantly, however, although participants did not show elicited

SCRs to the test stimuli, they still produced avoidance responses in their presence.

Given that in Experiment 2 all stimuli (either with direct or derived functions) were

tested, it can be concluded that avoidance behavior remained intact in the absence of

any noticeable increases in arousal (as measured by SCRs) for all participants but one.

This seems consistent with clinical literature that has regarded avoidance, rather than


33/60

32

fear, as the key component in the maintenance of anxiety disorders (e.g., Hayes, 1976;

Hayes et al., 1996; Forsyth et al., 2006; Powers, Smits, & Telch, 2004).

Equivalence Test

All participants but one (whose data were not relevant to the main analysis) who

were tested for equivalence passed successfully (see Appendix 2, Phase 6).

General Discussion

The main purpose of this study was to examine the impact of respondent

extinction on avoidance responding. Neither of the extinction protocols tested in

Experiments 1 and 2 made a significant impact on avoidance. Before addressing the

discussion of this finding, we will turn to commenting the results relating to the

acquisition and transfer of fear and avoidance.

Acquisition and transfer of fear and avoidance.

Conditional discrimination training. All 43 participants (18 from Experiment 1

and 25 from Experiment 2) met the conditional discrimination training criteria

established for the formation of two six-member classes. As with similar experiments

with smaller stimulus classes (e.g., Auguston & Dougher, 1997; Dymond, Roche,

Forsyth, Whelan, & Rhoden, 2007; Roche et al., 2008; Rodriguez-Valverde et al.,

2009), we observed substantial variability in the length of conditional discrimination

training. It is worth noting that despite the large number of relations participants had to

learn, training was relatively short. This was probably due to the use of an orderly

sequence of trials in which each new relation was trained only after the participant had

met the mastery criterion for the previous relation or set of relations being trained.

Conditioning and transfer of respondent elicitation and avoidance-evoking

functions. Taking the total amount of participants in both experiments as a whole, we


34/60

33

found that during conditioning 41 out of 43 participants met the avoidance/approach

criterion, and that 29 of those also met the respondent conditioning criterion. In

addition, a good number of participants (33 out of 41) showed transfer of

avoidance/approach functions to non-directly conditioned stimuli (that is, showed

avoidance to D1 or/and F1 and approach responses to D2 or/and F2), but only half of

them (16) showed transfer of respondent elicitation (i.e., showed larger SCRs to

D1/F1than to D2/F2). It is worth noting that the use of a differential conditioning

procedure wherein Class 1 stimuli acquired aversive functions while Class 2 stimuli

acquired appetitive functions, served methodological control purposes and was very

effective in establishing clearly distinct responses for each class during conditioning and

transfer. For instance, no participant produced approach responses to Class 1 stimuli,

and there were almost no avoidance responses to Class 2 stimuli; besides, only minimal

elicited SCRs to Class 2 stimuli were observed. Overall, these findings show that

avoidance responding is easily acquired and that avoidance-evoking functions easily

transfer to stimuli with no direct aversive conditioning history. More interestingly, they

show that avoidance responding may occur and transfer in the absence of noticeable

conditioned fear responses to the stimuli being avoided.

These results replicate previous findings on the transfer of respondent elicitation

and avoidance-evoking functions (e.g., Augustson & Dougher, 1997; Dougher et al.,

1994; Dymond et al., 2007; Roche et al., 2008; Rodrguez-Valverde et al., 2009) and

add to them in several respects. First, as in Auguston and Dougher (1997), a strict

criterion was established to determine whether transfer had occurred, namely,

performance on the first stimulus presentation of each class. The demonstration of

transfer during the first test trial with any stimulus is an important proof of derived

effects, because it is obtained prior to the participant experiencing any contingencies


35/60

34

(either experimentally implemented or inferred by the participant) with that very

stimulus. Second, transfer of functions was assessed before the equivalence test, which

rules out the impact of contiguity and second-order conditioning as potential

explanations for transfer effects (see Dymond & Rehfeldt, 2000).

The third and critical way these results add to previous findings relates to the

fact that this study included the simultaneous measurement of both respondent and

operant responses throughout the procedure. Although transfer of respondent extinction

and of avoidance-evoking functions is a well-known phenomenon now, this is the first

time that both types of responses were collected concurrently in the same experiment.

Besides, this preparation is relevant because of the interesting results it has yielded in

this regard, namely the lack of synchronicity between respondent elicitation and

avoidance responding. As mentioned previously, many participants learned to produce

avoidance responses to stimuli that did not elicit an increase in autonomic arousal (A

and B stimuli), and showed derived transfer of these avoidance responses to other

stimuli in an equivalence relation with the former (D and/or F), again in the absence of

any increase in elicited autonomic arousal to these stimuli. The main difference with

prior studies on the transfer of respondent elicitation (e.g. Dougher et al., 1994;

Rodrguez-Valverde et al., 2009) was that in the present study participants could

produce avoidance responses to prevent shock. Accordingly, this seems the most likely

and parsimonious explanation for the observed lack of elicited SCRs. In other words, it

might be the case that, for a significant number of participants, the eliciting functions of

the stimuli (during both conditioning and the transfer tests) were altered by the

availability of the opportunity to avoid. This is consistent with findings in the clinical

literature showing that the availability of opportunities to engage in avoidance (in the

form of safety behaviors) for patients with anxiety disorders may have a fear reduction


36/60

35

effect (e.g., Rachman, Radomsky, &Shafran, 2008). We agree with authors like De

Houwer, Vandorpe, and Beckers (2005), and Hermans et al. (2006), that the analysis of

verbal processes occurring during the experimental tasks (e.g., rules about the

occurrence of shock or about how to avoid it) is key for understanding conditioning in

humans. In our opinion, this sort of analysis may shed light on the reasons why

avoidance behavior had a fear reduction effect in the presence of shock-related cues

(cues that signaled impending shock). Propositional models like Lovibonds expectancy

theory of avoidance (Lovibond, 2006) seem a particularly appropriate option (from a

cognitive point of view) to explain these findings. Basically, during avoidance

conditioning participants would formulate the rule that shock will not be presented once

avoidance responding (that has been effective in the past) is produced. Accordingly,

presence of a shock-signaling stimulus would not elicit significant fear if the participant

were certain that they would have the opportunity to avoid (as was the case in these

experiments with the green context).

The effect of respondent extinction on avoidance behavior

As previously mentioned, our main goal was to examine the impact of

respondent extinction on both directly conditioned and derived avoidance, as an

exploratory analogue of exposure therapy. Accordingly, for the analysis of the impact of

respondent extinction, we only considered participants who (1) had shown differential

SCR conditioning and avoidance conditioning to A1/B1 (in Phase 2); and (2) had

produced derived avoidance responses to D1 and/or F1 (in Phase 3: transfer test).

In Experiment 1, respondent extinction in the same context in which aversive

conditioning took place did not prevent avoidance for 66.7% participants (6 out of 9). It

is important to notice that most of them (4/6) showed a recovery of previously


37/60

36

extinguished elicited SCRs in the context of avoidance. This fear-renewal effect has

been reported in the experimental and clinical literature on the context-sensitivity of

extinction (Craske & Mystkowski, 2006; Hermans et al., 2006; Neumann &

Longbottom, 2008; Vansteenwegen et al., 2006). The use of extinction in multiple

contexts has been presented as a way to reduce fear renewal and hence avoidance

(Hermans et al., 2006). In Experiment 2, we found that instructing participants that the

avoidance context was similar to the extinction context, served to significantly reduce

(almost eliminate) the fear-renewal effect (only one out of 10 participants showed

elicited SCRs to Class 1 stimuli when back to the avoidance context). Contrarily to

what is assumed, however, the reduction of elicited fear responses did not lead to less

avoidance: nine out of 10 participants continued producing avoidance responses to

Class 1 stimuli when back to the avoidance context. Apparently, context-similarity

training served to potentiate the lack of synchronicity between fear and avoidance

responses already observed in Experiment 1. Perhaps the participants expectations of

being shocked againas most of them indicated at the end of the study by informal post-

task reports played an important role (i.e., rules like: white and green are similar, but

pressing Q with the green circle works to prevent shock for sure, so Ill keep doing it

just in case). Again, we believe that this points to the potential relevance of further

research on the role of verbal processes (i.e., of rules such as expectations, beliefs, or

verbal formulation of past experiences) on avoidance conditioning and generalization.

In addition, the present findings have shown that trying to control avoidance

behavior just by extinguishing previously conditioned fear responses is a difficult

endeavor. Exposure can be conceived of as a strategy intended not only to the extinction

of conditioned fear responses, but also to disconfirming (i.e., changing the content of)

rules about avoidance in the feared situation or event (e.g., Now this is safe; nothing


38/60

37

bad really happens if I stand the situation) (e.g., Salkovskis, Hackmann, Wells, Gelder,

& Clark, 2007). In regard to this, we believe that it is worth considering the evidence

showing the futility (at least in some situations and for some individuals) of attempts to

changing the content of avoidance rules, and the potential utility of addressing the

function of those rules (as an intervention target) as the key for the maintenance of

pathological avoidance (Hayes, Wilson, et al., 1996). Accordingly, interventions aimed

at altering the believability of avoidance rules may well be an alternative to traditional

exposure-like techniques in targeting resistant forms of avoidance responding (see

Luciano et al., 2012, for an experimental analogue preparation).

We would like to point out some caveats in regard to the features of this study.

First, the participants expectations were not measured throughout the procedure, so no

information was obtained about what rules the participants were deriving and following

when they continued avoiding in the absence of noticeable fear responses. Although

concurrent measurement of expectancies has pros and cons (e.g., producing expectancy-

ratings on a trial by trial basis might interfere with physiological measures of

responding, or might have an effect on the very acquisition of conditioning), future

research might provide creative ways to target the symbolic nature of these experiences,

as indicated by De Houwer et al. (2005). Second, our preparation did not incorporate

specific conflicting contingencies for avoidance and non-avoidance other than the

presence or absence of shock. This was intended to make sure that avoidance was a

direct consequence of the aversive elicitation functions of Class 1 stimuli, and not

related to the loss of any additional short/long term appetitive consequences. However,

future studies might test whether the occurrence of additional contingencies for non-

avoidance enhances the effect of respondent extinction on avoidance responding. Third,

the conditional discrimination training procedure employed here, presented both the


39/60

38

sample and comparison stimuli simultaneously on the screen. It could be argued that

this might have resulted in unintended training of bidirectional relations, and thus that

equivalence occurred on the basis of directly trained symmetry. There is evidence,

however, that nonhumans and preverbal humans fail symmetry tests after conditional

discrimination training in the same format as that employed in the present series.

Additionally, no study has shown that the simultaneous presentation of sample and

comparison stimuli is a critical variable in determining whether or not derived relations

emerge during test trials. Finally, future studies might introduce a between-subjects

control condition in which participants were not presented with respondent extinction,

in order to gain further information about the role of such manipulation on subsequent

avoidance responding.

Conclusions

While exposure therapy is the treatment of choice for anxiety disorders (e.g.,

Deacon & Abramowitz, 2004), the processes that might be responsible of its

effectiveness are still in need of further analysis for a complete understanding.

Researchers have claimed for the need of analogue studies on how anxiety disorders are

developed and also on the mechanisms of the exposure techniques used to treat those

disorders (Abramowitz, in press; Craske & Mystkowski, 2006; Hermans et al., 2006).

Here we have presented translational research in which a commonly reported clinical

phenomenon, namely the use of exposure techniques and their effects on avoidance

behavior, was translated into basic processes to be tested in the laboratory context.

Specifically, with the present series of experiments we aimed at creating an analogue

that functionally matched, as closely as possible, a clinical situation in which the

therapist made in vivo exposure to the feared event in the same context where fear was


40/60

39

conditioned (white circle) but not in the context where avoidance behavior usually

occurs (green circle). In such hypothetical clinical situation, treatment would be over

once the patient successfully underwent successful extinction of conditioned fear

responses to the feared object and related events, and she could stay calm in the

conditioning/extinction context. However, what our results show is that as soon as the

patient is back to the context where it is possible to engage in avoidance responding

(green context), she continues to produce avoidance responses to the feared event and to

other stimuli in an equivalence relation with the former (Experiment 1). Possibly, one

common reaction by the clinician in such situation would be to work with the patient

and teach her that the exposure context and the context in which avoidance occurs do

not have to be different, that is, that just as she can stay calm in the exposure context,

she could stay calm in the avoidance one (i.e. the instruction the white and green circle

are similar in Experiment 2). But again, according to our results, although this seems to

alleviate anxiety (reduction of fear responses), the patient would continue to show

avoidance responding to the feared object when she had the chance to (i.e. the findings

in Experiment 2). Given that in this experimental series we have found a very little

effect of respondent extinction on avoidance, and in a attempt to further understand

what exposure techniques are made of, and under which conditions they work,

subsequent studies from our lab have incorporated a personal values context a

meaning and the practice of defusion from thoughts and sensations about impending

shock, in order to facilitate non-avoidance, with highly promising results (Luciano,

Valdivia, et al., 2012). We hope that this line of research will improve our

understanding of the process of change underlying the effectiveness of exposure

techniques for the treatment for anxiety disorders.


41/60

40


42/60

41

References

Abramowitz, J. S. (in press). The practice of exposure therapy: Relevance of cognitive-

behavioral and extinction theory.Behavior Therapy. doi:

10.1016/j.beth.2013.03.003

Augustson, E. M., & Dougher, M. J. (1997). The transfer of avoidance evoking

functions through stimulus equivalence classes.Journal of Behavior Therapy and

Experimental Psychiatry, 28, 181197.

Barlow, D. H. (2002).Anxiety and its disorders: The nature and treatment of anxiety

and panic. New York: Guilford Press.

Cohen, J. (1988). Statistical power analysis for the behavioral science (2nd

ed.).

Hillsdale, NJ: Erlbaum.

Craske, M. G., & Mystkowski, J. L. (2006). Exposure therapy and extinction: clinical

studies. In M. G. Craske, D. Hermans, & D. Vansteenwegen (Eds.),Fear and

learning: From basic processes to clinical implications (pp. 217-233).

Washington DC: American Psychological Association.

Dawson, M. E., Schell, A. M., & Filion, D. L. (1990). The electrodermal system. In J.

T. Cacioppo, & L. G. Tassinary (Eds.),Principles of psychophysiology (pp. 295-

324). New York: Cambridge University Press.

De Houwer, J. (2009). The propositional approach to associative learning as an

alternative for association formation models.Learning & Behavior, 37, 1-20.

De Houwer, J., Vandorpe, S., & Beckers, T. (2005). On the role of controlled cognitive

processes in human associative learning. In A. J. Wills (Ed.),New directions in

human associative learning(pp. 41-63). Hillsdale, NJ: Lawrence Erlbaum

Associates.


43/60

42

Deacon, B. J., & Abramowitz, J. S. (2004).Cognitive and behavioral treatments of

anxiety disorders: a review of meta-analytic findings.Journal of Clinical

Psychology, 60, 429-441.

Dougher, M. J., Augustson, E., Markham, M. R., Greenway, D. E., &Wulfert, E.

(1994). The transfer of respondent eliciting and extinction functions through

stimulus equivalence classes. Journal of the Experimental Analysis of Behavior,

62, 331351.

Dougher, M. J., Hamilton, D. A., Fink, B. C., & Harrington, J. (2007). Transformation

of the discriminative and eliciting functions of generalized relational stimuli.

Journal of the Experimental Analysis of Behavior, 88, 179-197.

Dymond, S., & Rehfeldt, R. A. (2000). Understanding complex behavior: The

transformation of stimulus functions. The Behavior Analyst, 23, 239-254.

Dymond, S., & Roche, B. (2009).A contemporary behavior analysis of anxiety and

avoidance.The Behavior Analyst, 32, 7-27.

Dymond, S., Roche, B., Forsyth, J. P., Whelan, R., & Rhoden, J. (2007).

Transformation of avoidance response functions in accordance with same and

opposite relational frames.Journal of the Experimental Analysis of Behavior, 88,

249-262.

Forsyth, J. P., Eifert, G. H., & Barrios, V. (2006). Fear conditioning in an emotion

regulation context: A fresh perspective on the origins of anxiety disorders. In M.

G. Craske, D. Hermans, & D. Vansteenwegen (Eds.),Fear and learning: From

basic processes to clinical applications (pp. 133-153). Washington, DC:

American Psychological Association.

Hayes, S. C. (1976). The role of approach contingencies in phobic behavior.Behavior

Therapy, 7, 28-36.


44/60

43

Hayes, S. C., Wilson, K. G., Gifford, E. V., Follette, V. M., & Strosahl, K. D. (1996).

Experiential avoidance and behavioral disorders: a functional dimensional

approach to diagnosis and treatment.Journal of Consulting and Clinical

Psychology, 64, 1152-1168.

Hermans, D., Craske, M. G., Mineka, S., & Lovibond, P. F. (2006). Extinction in

Human Fear Conditioning.Biological Psychiatry, 60, 361-368.

Lovibond, P. F. (2006). Fear and avoidance: An integrated expectancy model. In M. G.

Craske, D. Hermans, & D. Vansteenwegen (Eds.),Fear and learning: Basic

science to clinical application (pp. 117-132). Washington, DC: American

Psychological Association.

Lovibond, P. F., Davis, N. R., & OFlaherty, A. S. (2000). Protection from extinction in

human fear conditioning.Behaviour Research and Therapy, 38, 967-983.

Lovibond, P. F., Mitchell, C. J., Minard, E., Brady, A., &Menzies, R. G. (2009). Safety

behaviours preserve threat beliefs: Protection from extinction of human fear

conditioning by an avoidance response.Behaviour Research and Therapy, 47,

716-720.

Luciano, C., Valdivia, S., Ruiz, F. J., Rodriguez, M., Barnes-Holmes, D., Dougher, M.

J., et al. (2012). A laboratory demonstration of the benefits of using a defusion-

based intervention to reduce avoidance behavior.Paper submitted for

publication.

Marks, I. M. (1981). Cure and care of neuroses. Wiley: New York

Mowrer, O. H. (1960).Learning theory and behavior. New York: Wiley.

Neumann, D. L., & Longbottom, P. L. (2008). The renewal of extinguished conditioned

fear with fear-relevant and fear-irrelevant stimuli by a context change after

extinction.Behaviour Research and Therapy, 46, 188-206.


45/60

44

Powers, M. B., Smits, J. A. J., & Telch, M. J. (2004). Disentangling the effecs of safety-

behavior utilization and safety-behavior availability during exposure-based

treatments: A placebo-controlled trial.Journal of Consulting Clinical Psychology,

72, 448-454.

Rachman, S. (1977). The conditioning theory of fear-acquisition: a critical examination.

Behaviour Research and Therapy, 15, 375-387.

Rachman, S. (1991).Neo-conditioning and the classical theory of fear

acquisition.Clinical Psychology Review, 11, 155-175.

Rachman, S., Radomsky, A. S., & Shafran, R. (2008). Safety behavior: A

reconsideration.Behaviour Research and Therapy, 46, 163-173.

Roche, B., Kanter, J. W., Brown, K. R., Dymond, S., Fogarty, C. C. (2008). A

comparison of direct versus derived extinction of avoidance. The

Psychological Record, 58, 443-464.

Rodrguez-Valverde, M., Luciano, C., & Barnes-Holmes, D. (2009). Transfer of

aversive respondent elicitation in accordance with equivalence relations.Journal

of the Experimental Analysis of Behavior, 92, 85-112.

Salkovskis, P. M., Hackmann, A., Wells, A., Gelder, M. G., & Clark, D. M. (2007).

Documents

ACT. Extinction of Aversive Eliciting Functions as an. Luciano_et_al_in_press