Performance during extinction as a function of the number of reinforcers delivered during training

MISSING IMAGE

Material Information

Title:
Performance during extinction as a function of the number of reinforcers delivered during training
Physical Description:
v, 57 leaves : ill. ; 29 cm.
Language:
English
Creator:
Zarcone, Troy J., 1965-
Publication Date:

Subjects

Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1994.
Bibliography:
Includes bibliographical references (leaves 53-56).
Statement of Responsibility:
by Troy J. Zarcone.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002054622
notis - AKP2602
oclc - 33486668
System ID:
AA00003647:00001

Full Text









PERFORMANCE DURING EXTINCTION AS A FUNCTION OF THE
NUMBER OF REINFORCERS DELIVERED DURING TRAINING



By

TROY J. ZARCONE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY



UNIVERSITY OF FLORIDA


1994












ACKNOWLEDGE IENTS

I thank my wife, Jennifer Zarcone, for introducing me to the science of behavior

and for providing me with love, support, and a family (i.e., Nicholas Zarcone). I thank

my mother, Gail Scribner, for teaching me the joy of learning and for her love and support.

I also thank my advisor, Henry Pennypacker, for giving me the chance to do behavioral

research and for his friendship and guidance. I wish to thank Marc Branch for the use of

him and his facilities in my endeavor to become a student of behavior. Special thanks go to

Christine Hughes, for her patience in teaching me how to do behavioral research. I thank

my committee, Marc Branch, Tim Hackenberg, Frans Van Haaren, Brian Iwata, Ed

Malagodi, Cecil Mercer, Henry Pennypacker and Don Stehouwer, for their time and effort

in the review and preparation of this dissertation.












TABLE OF CONTENTS


ACKNOWLEDGMENTS...................................................................... ii

ABSTRACT ....................................................................... ...... .... .... iv

INTRODUCTION ............................................... ............ ................... 1

M ETH OD ......... ......... ....................................... ................................ .. 8
Subjects................................. .................................... 8
A apparatus ..................................................... .. .... ...... ............ .... .... 8
P rocedure........................................ .............................. ......... 9
Stability Criteria ................... .................................. ... .... ............ 13
Extinction Criterion ........................................................ ...........15
RESU LTS..................................................................16

DISCUSSION ........... ..... ....... .......... ...... ........ .. ... ............... ............. 41

APPENDIX A: COMPONENT DURATIONS (MIN) DURING EXTINCTION.........50

APPENDIX B: STABILITY AND VARIABILITY STATISTICS OF THE VR VR AND
VR CRF PHASES ......................................... .............. ....................... 52

REFERENCES................................................... ..............................53

BIOGRAPHICAL SKETCH....................................................................57












Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


PERFORMANCE DURING EXTINCTION AS A FUNCTION OF
THE NUMBER OF REINFORCERS DELIVERED DURING TRAINING.


By


Troy J. Zarcone

December 1994


Chairman: Dr. H. S. Pennypacker
Major Department: Psychology

Keypecking by seven pigeons was established and maintained on a multiple

variable ratio (VR), VR schedule of food presentation. The schedule in one of the

components was then changed to continuous reinforcement (CRF) for a predetermined

number of reinforcers. Both components were then changed to extinction (i.e., multiple

extinction, extinction). This sequence was repeated for a varied number of times for each

pigeon in order to determine the relationship between the number of reinforcers delivered

during each component of the multiple VR, CRF schedule and responding during

extinction. For most pigeons there were fewer responses during extinction in the presence

of a stimulus recently paired with CRF, regardless of the number of reinforcers received in

that component. The ratio of the total responses in extinction in the former VR

reinforcement component to the total responses in the CRF component increased as the

number of reinforcers delivered during each component of the multiple schedule increased.

This ratio measure of the extinction performances provided an estimate of the magnitude of

the partial reinforcement extinction effect (PREE) which was reproduced consistently both






between and within subjects. Within-subject replications of the PREE generally occurred

without an overall reduction in the number of responses with repeated exposures to

extinction. The rate of change in response rate during extinction was highly variable within

and across subjects when the multiple VR, CRF exposure was less than 1500 reinforcers

per component. Exposure to more than 1500 reinforcers per component during the

multiple VR, CRF schedule resulted in roughly equal rates of change for the two

components. These data confirm the existence of a PREE in individual pigeons and that the

effect can be reproduced within subjects. Further, these data suggest that with sufficient

training between exposures to extinction, the repeated-extinction effect can be minimized or

eliminated.












INTRODUCTION

The "persistence" of behavior after the discontinuation of reinforcement is

considered important to the survival of an organism in an ever-changing environment

(Skinner, 1938). In a world in which access to food, water, and sexual partners does not

always occur reliably, organisms that continue to respond after several unsuccessful

attempts at obtaining these commodities may have a selective advantage compared to

organisms who "give up" if the persistence eventually leads to obtaining the required

commodity. Persistent responding when a resource has been used up (e.g., over fishing a

small pond) could also be detrimental if it prevented the organism from going on to more

profitable activities (e.g., searching a bush for berries). Any general theory of behavior

must be able to explain the persistence or lack of persistence in behavior when

contingencies for that behavior have changed. Of equal importance is that an understanding

of the variables that affect persistence will serve as a powerful tool in the application of a

behavioral technology. The present experiment is an analysis of a procedure designed to

examine the effect of some of the possible controlling variables in the persistence of

behavior in a single organism.

The usual method of examining the effects of intermittent reinforcement on

persistence is the between-group comparison (see Jenkins & Stanley, 1950, for a review).

Separate groups of subjects are exposed to different schedules of reinforcement which are

then followed by the withdrawal of reinforcement. The usual result is a robust effect called

the partial reinforcement extinction effect (PREE). The PREE describes the greater

persistence in responding during extinction following a history of partial, or intermittent,

reinforcement (PRF) compared to a history in which each response is reinforced or

continuous reinforcement (CRF). The simplest schedule of reinforcement in an






experimental setting is CRF, which specifies that a reinforcer is presented after every

instance of a specified operant response class. More complex schedules are referred to as

intermittent or partial schedules under which reinforcers are delivered after only some

responses. Extinction (EXT) of operant behavior is the procedure of discontinuing

reinforcement of a specified operant response class (Iversen & Lattal, 1991). The term is

also used to describe the process of the reduction in the frequency of the operant response

class resulting from the extinction procedure (Iversen & Lattal, 1991).

Use of separate groups of subjects usually produces the PREE, but attempts to

generate the PREE in individual subjects have met with difficulty. One method of

comparing the effects of continuous and intermittent reinforcement histories within an

organism is to expose that organism to both CRF and PRF at different times during the

acquisition procedure. Each schedule of reinforcement can be presented with a different

discriminative stimulus. The different discriminative stimuli are then presented at different

times without reinforcement during the extinction phase of the procedure.

Many of the previous attempts at generating a within-subject PREE have produced

reversed PREEs, more persistent responding after CRF compared to PRF (Adams,

Nemeth, & Pavlik, 1982; Nevin, Mandell & Atak, 1983; Pavlik & Carlton, 1965; Pavlik,

Carlton, Lehr, & Henrickson, 1967). One possible explanation for this inconsistency in

the results of between-subject and within-subject procedures involves the practice of

maintaining equal temporal exposure to discriminative stimuli associated with PRF and

CRF, but at the same time presenting fewer reinforcers under the PRF schedule compared

to the CRF schedule (e.g., Adams, Nemeth, & Pavlik, 1982; Nevin, Mandell, & Atak,

1983; Pavlik & Carlton, 1965; Pavlik, Carlton, Lehr, & Henrickson, 1967). The present

experiment was an attempt to produce the PREE within a single organism while keeping the

number of reinforcers delivered during PRF and CRF equal.

For a within-subject procedure to succeed in producing the PREE, stimuli paired

with CRF and PRF during training must control responding. The lack of stimulus control






between CRF and PRF conditions would produce an effect on extinction responding akin

to extinction after a single PRF schedule. A possible example of a failure of stimulus

control can be seen in Rashotte, Ross, and Amsel's (1968) assessment of the effect of the

overall rate of reinforcement for a within-subject comparison. Three groups of rats were

exposed to two runways, one painted black on the left and white on the right and the other

painted in the reverse pattern. Each runway was assigned one of three schedules of

reinforcement. One group of subjects received reinforcers on 50% of the trials in one

runway and 100% of the trials in the other runway (50/100 group). Another group of

subjects received reinforcers on 50% of the trials in both runways (50/50 group). The last

group of subjects received reinforcers on 75% of the trials in both runways (75/75 group).

The 50/100 group and the 75/75 group received the same overall rate of reinforcement of

75%, whereas the 50/50 group received an overall rate of reinforcement of 50%. No

differences were seen between the mean alley speeds for the two runways in the last

acquisition block (i.e., 2 trials in each alley) for any group. All groups showed no

differences in reduction in mean alley speeds during extinction between the two runways.

Comparisons among the groups revealed that the speeds during extinction of the 50/50

group (i.e., 50% overall) took longer to decrease than those of the 50/100 and 75/75

groups (i.e., 75% overall), which decreased at a similar rate. These results may be

interpreted as a lack of stimulus control by the stimuli associated with the different

schedules of reinforcement. The 50/100 group was a standard within-subject PREE

procedure but it produced results similar to the 75/75 group due to a lack of stimulus

control by the runway paint pattern for the individual schedules of reinforcement.

Several experimenters have attempted to generate a within-subject PREE by

accentuating the stimulus differences associated with the CRF and PRF schedules.

Rashotte (1968) required rats to emit the same response (running) for both schedules or

different responses (running and climbing) for each schedule. He found a PREE if the

responses were the same, but the magnitude of the PREE was increased by requiring






different responses for each schedule. Amsel, Rashotte, and MacKinnon (1966) used

different colored runways (black vs. white) and were unable to produce a PREE when

equating the number of reinforcers (90) or the number of trials (84) during training. They

did, however, produce a small PREE after exposing subjects to 160 trials during training.

Rashotte and Amsel (1968) attempted to increase stimulus control with either a black or

white runway and exposure to the trials (60) during either the morning or evening, but no

PREE was produced. Waters and Knott (1970) used either barehanded vs. glove handled,

noise vs. silence, morning vs. evening, and the construction of the runways as

discriminative stimuli and succeeded in producing a within-subject PREE. Feider (1973)

reported that groups of rats required to respond on different levers showed a PREE,

whereas groups that had to responded on one lever showed no PREE.

In light of these experiments, particular attention must be paid to the stimuli used in

conjunction with the schedules of reinforcement if a within-subject PREE is to be

reproduced. The preceding experiments concentrated on increasing the discriminability of

stimuli associated with the different schedules of reinforcement. Another strategy is to take

advantage of an organism's susceptibility to stimulus control by a single stimulus modality

(e.g., light wave frequency in the case of pigeons, Terrace, 1966). Using a single

response in the presence of two different stimuli of the same modality reduces the possible

confounds of comparing topographically different responses (e.g., differences in effort or

time to complete the responses).

In the present experiment, an attempt was made to determine the function relating

the number of CRF reinforcers presented after PRF and before EXT (i.e., interpolated

CRF) to the number of responses made during EXT to determine if interpolating a period

of CRF between exposures to intermittent reinforcement and extinction would accelerate

extinction. If PRF produces more persistence during extinction then CRF, implementing a

schedule of CRF after PRF just before attempting to extinguish the behavior may reduce

persistence of responding characteristic of intermittently reinforced behavior.







Results from early experiments examining the effects of combinations of CRF and

PRF on extinction performances have been mixed. Many of the earlier experiments

showed no reductions in extinction performances with the introduction of CRF after PRF

(Jenkins, 1962; Likely, 1958; Quartermain & Vaughan, 1961; Sutherland, Mackintosh &

Wolfe, 1965; Theios, 1962), but other experiments did produce a reduction in persistence

(Dyal & Sytsma, 1976; Moreland, Stalling & Walker, 1983; Stalling, Moreland, Merrill &

Scotti, 1981).

A possible explanation for the differing results may lie in the method used to assess

the effects of interpolated CRF on extinction performances compared to extinction

performances after PRF alone. For example, Jenkins (1962, Experiment 2) compared a

group of pigeons receiving 20 PRF (50%) trials to a group receiving 20 PRF followed by

12 CRF trials, and showed no significant difference. In this comparison, the number of

PRF trials was equal (20) but the total number of trials and reinforcers were not equal (20

vs. 32 trials and 10 vs. 24 reinforcers). Comparing the 20-PRF group to another group

that received 8-PRF-followed-by-12-CRF, thereby equating the number of trials, showed a

decrease in responding during extinction when CRF followed PRF. Comparison of the

32-PRF group to the 8-PRF-followed-by-12-CRF group, which equated the number of

reinforcers, also showed a decrease in extinction responding with interpolated CRF.

Jenkins' (1962) data, along with data from other experiments (e.g., Hearst, 1961;

Perin, 1942; Williams, 1938; Wilson, 1954) have shown that the number of reinforcers

delivered during training can have a pronounced effect upon extinction performance. In

order to compare the effects of CRF and PRF under equal conditions and increase the

likelihood of producing a PREE, the present experiment was designed to keep the number

of reinforcers presented during CRF and PRF equal.

To approximate conditions often observed in natural settings, a variable ratio (VR)

schedule was selected for the PRF schedule. Variable-ratio schedules are simple in

comparison to the many possible intermittent reinforcement schedules found in nature, but







the strategy of the present experiment was to create an intermittent schedule of

reinforcement under controlled conditions without introducing such additional variables as

noncontingent reinforcement or interresponse time requirements.

An additional goal of the present experiment was to evaluate the use of different

extinction criteria in establishing the existence of the PREE. An important procedural

consideration is the decision of when to stop measuring behavior during extinction. The

extinction criterion chosen should allow for a complete picture of extinction responding

without unduly lengthening the experiment. The simplest extinction criterion has been to

conclude the procedure after a predetermined duration (e.g., 5 daily 40 min. sessions,

Margulies, 1961; or 2 daily 20 min. sessions, Hearst, 1961). Another fairly simple

criterion is to expose the subject to extinction until a predetermined amount of time without

a response has occurred (e.g., 1 session until 5 min. without a response, Perin, 1942;

Williams, 1938). The time-without-a-response criterion could also be used with repeated

daily sessions, for example, 5-min daily sessions with a 5-min-without-a-response

criterion (Hothersall, 1966) or more simply, one session without a response. This

experiment employed a conservative extinction criterion, thus allowing for retrospective

analyses of other less stringent extinction criteria.

Nevin (1974) introduced a measure of response strength that equated response

measures generated during different reinforcement schedules. With respect to this

measure, the schedule with the higher rate of reinforcement generally produced behavior

that was more resistant to environmental changes that followed. Nevin's theory of

behavioral momentum was an analogy drawn from physics that emphasized the relationship

between external variables, response strength and resistance to change. Behavioral mass

(i.e., response strength) was inferred from relative changes in response rate when

environmental conditions are changed (Nevin, Mandell, & Atak, 1983). Nevin's theory

predicts that CRF would produce more resistance to change compared to PRF (i.e., the

reversed PREE). The reversed PREE would be predicted because the rate of reinforcement





7

for CRF is usually higher than for PRF and the higher the rate of reinforcement, the more

resistant behavior becomes to environmental changes (e.g., extinction). An additional

function of the present experiment was to examine the rate of change in extinction

responding as this measure was affected by the number of CRF reinforcers.














METHOD


Subjects

Seven adult, experimentally naive, male White Carneau pigeons (Columba livia)

were housed individually in a colony room (16 hr. light/8 hr. dark). They had continuous

access to vitamin enriched water and health grit. Subjects 2374, 3180 and 3519 started the

experiment at 80% of their free-feeding weights, but were later reduced to 70% to increase

stability of responding (see below). Subjects 2822, 2842, 2875 and 2880 were maintained

at 70% of their free-feeding weights throughout the experiment. Each bird was fed after

the experimental session as necessary to maintain the desired body-weight.

Apparatus

The operant conditioning chamber was a Lehigh Valley Electronics (Model 1519c)

unit for pigeons with workspace dimensions of 35.5 cm x 30.5 cm x 35.5 cm. The

chamber was fitted with a three-key display board (Model 132-02). Three response keys,

2.5 cm in diameter, were located 5.6 cm from each other and 23 cm from the chamber floor

in a horizontal row on the display wall. Only the center key was used in the experiment. A

keypeck with a force greater than 0.18N started a 70-ms tone from a Sonalert (Model

SC628). The Sonalert was located behind the front wall 2.0 cm from the floor. The

center key could be transilluminated from the rear by a white, green (G) or red (R) light.

The left key was dark and inoperable and the right key was replaced with an aluminum

plate. The chamber was illuminated by a 1.2-W bulb (houselight) 5.5 cm above the center

key.

During operation of the food hopper, pigeons gained access to mixed grain (5/10

milo, 4/10 buckwheat, 1/10 hemp). The opening to the food hopper was 5.7 by 5.2 cm and






located 9.0 cm below the center key. The inside of the food hopper was illuminated by a

1.2-W bulb and the keylight and houselight were turned off during food presentations.

The chamber was kept in a room where white noise was continuously present. A

custom-built computer (Walter & Palya, 1986) was mounted in a metal enclosure on top of

the chamber. The computer operated under the ECBasic control system (Walter & Palya,

1986) that programmed contingencies, collected data and interfaced with an IBM-

compatible computer (Zenith) in an adjacent room. A Gerbrands (Model C-3) cumulative

response recorder was used to monitor responding.

Procedure

Each pigeon was placed in the chamber for at least two 30-min adaptation sessions

in which the houselight was on and no behavioral contingencies were programmed. Once a

pigeon was moving around the chamber consistently, food hopper (magazine) training

began. Initially, the houselight was off and the hopper was raised and filled with mixed

grain. Once a pigeon ate from the hopper for 10-15 s the hopper was lowered (the

houselight turned on) and raised quickly. The hopper-presentation duration was gradually

shortened to 3 s, and the inter-food interval was lengthened to an average of 1 min.

Training continued until pigeons would reliably (and with short latencies) approach and eat

from the hopper from anywhere in the chamber.

Keypecking was shaped by differentially reinforcing successive approximations to

the final response. The key was illuminated by a white light during this initial training of

the keypeck. After keypecking was maintained for two sessions of CRF (40 reinforcers

per session) the schedule was changed to a multiple CRF (green keylight), CRF (red

keylight) schedule in which the components were presented in simple alternation, lasted for

5 reinforcers, and were separated by a one minute time-out (houselights and keylights

turned off) between the components. The response requirement was increased gradually

until keypecking was maintained on two equal VR schedules for both the green and red

keylights. Table 1 shows the terminal VR schedules for each pigeon. The VR schedules






were constructed by randomly selecting, without replacement, values from an evenly

distributed list of 25 numbers ranging from 1 to twice the value of the VR schedule.

Variable ratio values of 50 and 25 were used for some pigeons to maintain responding

without pauses after attempts at VR100 failed to do so consistently.



Table 1.
Schedules, keylight color assignments and number of components per session.
Green Red Number of
keylight keylight components/
Pigeon component component session
2822 VR100 VR100 or CRF 8
2875-1 VR100 VRI00 or CRF 8
2875-2 VR25 VR25 or CRF 8
2875-3 VR25 VR25 or CRF 8
2842 VR50 VR50 or CRF 8
3519 VR50 VR50 or CRF 10
2374 VR100 VR100 or CRF 10
3180 VR100 or CRF VR100 10
2880 VR50 or CRF VR50 8
Note: For subject 2875 different VR schedules were used between the first extinction series and the
extinction series that followed (i.e.. 2 & 3).



The experiment consisted of repeated exposures to a three-phase series. In the first

phase of the experiment reinforcers were arranged according to a multiple VR VR schedule

(VR VR). When stability criteria were met (see below) the second phase was implemented

using a multiple VR CRF schedule of reinforcement (VR CRF). For the majority of the

pigeons the VR CRF phase continued until a predetermined number of reinforcers was

delivered. The VR CRF phase for Pigeons 2374, series 1 and 2, 3519, series 1 and 2, and

3180, series 1, was terminated when performances had reached a steady state. The third

phase consisted of a multiple extinction, extinction (EXT EXT) schedule and was

terminated after no keypecking occurred for 5 daily consecutive sessions. Table 2 shows

the number of sessions for each phase of the extinction series and the number of reinforcers

for phases VR VR and VR CRF. The numbers listed below each pigeon's identification

number indicates the extinction series in the order in which it was presented to the pigeon.






Due to experimental errors (exposure to VR VR or VR CRF schedules during

extinction), EXT EXT phases for 2842-2, 3180-3, 3519-2,3,4 and 6 were not continued to

the extinction criterion and are denoted by an asterisk (*). An equipment failure caused a

dimmer green keylight to be presented during VR CRF and EXT EXT of extinction series 1

for Pigeon 2880 and extinction series 4 for Pigeon 3519. This keylight stimulus change

occurred 13 and 24 sessions before changing to EXT EXT, respectively. These extinction

series are denoted by a "+." The data for subjects exposed to experimental errors are

presented for the individual subjects, but are not included in the group analyses.

The starting component of the multiple schedule for each phase was determined

each day by a Gellermann (1933) series (G, R, G, G, R, G, R, R; repeats). Individual

components ended after 5 reinforcer presentations and the session ended after 10 (5 of each

keylight color) or 8 (4 of each keylight color) components (see Table 1).

The duration of each component during EXT EXT (Appendix A) was calculated in

the following manner. A geometric mean of the duration of each individual component

(e.g., 5 red and 5 green) was calculated from the last five sessions under the VR

schedule. For some extinction series, the individual geometric means for each of the

components (e.g., 5 green and 5 red) were used to time the components during extinction

(see Appendix A). In other instances the component durations during extinction were all

the same for each color (green and red) and were based on the arithmetic average across the

session for each keylight color in the last five sessions under the VR schedule. In the

remaining cases, the longer of the two arithmetic means of the component durations was

used for all components during extinction (see Appendix A). The rationale for the first two

timing procedures was done to minimize the differences between VR CRF and EXT EXT

while maintaining relatively equal exposure to both stimuli during extinction. In all EXT

EXT phases (except 2374-1), the two components of the multiple schedule were either

presented for equal durations or the stimulus previously paired with CRF was presented for







a longer time (see Appendix A). Instances of longer exposure to the CRF discriminative

stimulus during extinction were done to provide a conservative estimation of the PREE.



Table 2.
Number of sessions in each phase, number of reinforcers presented during the VR VR and
VR CRF phases, and total number of responses during extinction after each schedule.


Total Reinforcers Extinction
Number of Sessions Per Component Responses After
Pigeon VR VRI VR CRF I EXT EXT VR VR I VR CRF VR I CRF


1350
950


220
600


1220
760
440
600


1540
160
600


2480
580
1800


6500
1800
1725
600
450
750


* did not meet 5 session extinction criterion
+ change in green keylight in VR CRF phase.


2822
1
2

2842
1
*2
3
4

2875
1
2
3

2880
+1
2
3

3180
1
2
*3
4
5
6

3519
1
*2
*3
+*4
5
*6
7


2374
1
2


4767
1814


1805
1223


11763 10927
5222 5056


3250
1550


580
20


900
2020
3000
40


580
2500
20


1820
220
1260


2250
575
1975
1225
25
0


1250
300
3225
2250
1225
25
0


1952
1371
532
1896


2129
1310
2823


417
3145
2973


5202
9372
9044
8535
7358
3193


5321
4185
1841
1933
1987
1708
2987


2979
1809
2447
1859


3470
3622
1453


3335
4419
4398


10371
14267
20010
10275
8899
4331


7292
5816
4608
3400
3642
2979
3251


1000
1250
425
475
350
175
375






After responding extinguished during the EXT EXT schedule, the subject was

exposed to a multiple CRF CRF schedule during the following daily session. The schedule

of reinforcement was then adjusted back to the multiple VR VR schedule designated in

Table 1. Retraining consisted of incrementing both VR schedules (e.g., CRF, VR2, VR5,

VR10, VR20, VR50, VR75, VR100) when the daily cumulative records showed consistent

responding without pauses.

If a pigeon did not respond within approximately 5 min. after the beginning of the

multiple CRF CRF session, one of two priming procedures was used to generate

responding. The first procedure consisted of operating the food hopper once or twice for

3s independent of responding. The second procedure consisted of the experimenter

opening the chamber door and pressing the lighted center key with his finger. A few

pigeons began keypecking within the first 5 min. of the multiple CRF CRF and the rest of

the pigeons began keypecking after one of the two priming procedures.

To decrease variability in responding for Pigeon 3180, the hopper duration was

changed from 3s to 2.5s to reduce possible satiation effects towards the end of daily

sessions. This change in the hopper duration occurred before the first extinction series.

Stability Criteria

Stability criteria were employed to determine when to change phases at several

points in the experiment. Stability criteria were calculated using kappa (K) (Johnston &

Pennypacker, 1980) and celebration (C) (Lindsley, 1969; cited by Johnston & Pennypacker,

1980). Celeration is computed as the ratio of two predicted daily response frequencies

separated by seven days and is based upon the slope of the least squares regression lines

for the last 5 sessions of a phase. The unit of celebration value is movements (i.e.,

keypecks)/minute/week. A value of 1.00 indicates no change in trend. Values greater than

1.00 designate an increasing trend, and values less than 1.00 indicate a decreasing trend.

Values of celebration can range from 0 to infinity. The greater the change in value of

celebration from 1.00, the greater the increase or decrease in the rate of change in the




14

response rate measure. Celeration is an estimate of the trend in the average daily response

rates for each component and was used in conjunction with kappa to determine steady state

performance.

n

The computation formula for kappa is: antilog 2 I In Xi -In Xj In
I n(n-1)

For any data set, therefore, kappa may be regarded as the geometric mean of all ratios

(Xi/Xj) where Xi is greater than Xj. Kappa quantifies the variability seen in the repeated

measures of behavior independent of the mean of the data sample. This characteristic of

kappa makes for a more accurate comparison of variability between low and high response

rates (e.g., response rates produced by CRF and VR100 respectively). Values of kappa

range from 1.00 to infinity. The larger the value of kappa, the greater the variability in the

data sample. A kappa value of 1.10, for example, indicates that each value in the set

deviates from every other value, on average, by 10%.

The stability criteria were arbitrarily defined as celebration values between 0.90 and

1.10 and kappa values less than or equal to 1.10 for each component (green or red) of the

multiple schedule. Measurements of kappa and celebration were based on the response rates

for each component during the five most recent consecutive daily sessions.

In the VR VR phase an additional stability criterion, the Geometric Mean Ratio

(GMR), measured the equality of responding between the two VR components and was

computed as the ratio of the geometric means (GM) of the response rates for the last 5

sessions for each component, the larger geometric mean divided by the smaller. The value

of the GMR could range from 1.00 to infinity. The larger the value of the GMR, the

greater the difference between the two components. Equivalence between the two

components was defined as a GMR value less than or equal to 1.10. The values of kappa,

celebration, and GMR for each extinction series are shown in Appendix B.

For the last extinction series replication for Pigeons 2822, 2842, 3875, 2880 and

3180, changes from VR VR to VR CRF occurred after approximately 30 sessions





15

independent of the stability criteria. Due to the focus of the experiment on the manipulation

of the number of reinforcers delivered during VR CRF, most phase changes to EXT EXT

had to be made independent of the above mentioned stability criteria (see Appendix B).



Extinction Criterion

The EXT EXT phase was terminated after 5 consecutive daily sessions with no

responding in either component.












RESULTS

Figure 1 shows a sample of the cumulative records for Pigeon 2875's first

extinction series. This series was randomly chosen from those series showing the PREE.

Keypecking maintained under the VR100 (Figure 1, upper left record) showed consistently

high rates with occasional short pauses following reinforcement. Variable-ratio

performances in the green keylight component (event pen deflected down) remained

unaffected by the change in the red component (event pen deflected up) from a VR 100

schedule to a CRF schedule (Figure 1, upper right record). The lower left record (Figure

1) shows that the rate of keypecking was high for the first four components with an abrupt

change to almost no keypecking in the fifth (green keylight) component. Keypecking

remained at a near zero rate for the sixth component, but high rates of keypecking occurred

again throughout the seventh component (green keylight) only to return to near zero for the

final component of the session. On the second day (Figure 1, lower right record), the

session started with the red keylight component and keypecking rates were high for the first

two components. Keypecks occasionally occurred during the third through the seventh

component, with high rates of keypecking occurring again, halfway through the last

component (green keylight). The rate of keypecking differed from previous components in

that responding took on more of a "break and run" characteristic seen when keypecking

was being reinforced on the VR 100 schedule.

The last two columns in Table 2' show the number of responses to the extinction

criterion made in the presence of the two discriminative stimuli (i.e., red or green keylight)

previously paired most recently with VR or CRF. In 25 of 27 cases, there were fewer total



'Data from both extinction curves from 2374 and the first extinction curve from 3180 were published in a
master's thesis by the author.




17

responses during extinction after a recent CRF history than after a VR history. Continuous

reinforcement on two occasions produced more total responses during extinction when

CRF training was limited to 40 (2842-4) or 20 (2875-3) reinforcers.

Figures 2 through 5 show cumulative extinction curves for all seven pigeons.

For the first few sessions of extinction there was a high rate of responding after both CRF

and VR schedules. The VR history generally would occasion a high rate of responding for

a few additional sessions before tapering off. Significant differences generally were not

seen between extinction components after the first 10 to 15 sessions of extinction.

Responding during extinction after CRF occasionally would temporarily exceed

responding after VR (i.e., Figure 2, panel 3 (2875), Figure 3, panel 4 (2842); Figure 5,

panel 1, 2 and 5 (3519)), but this effect was sustained in only two instances (Figure 2,

panel 3 (2875), Figure 3, panel 4 (2842)). Pigeons 2875 and 2842 showed higher

numbers of responses after CRF compared to VR reinforcement after receiving relatively

few reinforcers during the previous VR CRF phase (20 and 40, respectively).

Figure 6 shows the total keypecks emitted during extinction for the CRF history

component (upper panel) and for the VR history (middle panel) as a function of the

presentation order of the extinction series. The lower panel of Figure 6 shows the

extinction ratio (i.e., total responses emitted during extinction in the presence of the VR

discriminative stimulus divided by the total number of responses emitted during extinction

in the presence of the CRF discriminative stimulus) as a function of the presentation order

of the extinction series. An extinction ratio value of 1.00 indicates no difference between

the extinction performances after VR and CRF. Extinction ratio values greater than 1.00

indicate a PREE. Values less than 1.00 indicate the opposite effect, a reversed PREE. The

larger the absolute value of the extinction ratio the greater the differences in extinction

performances after VR and CRF. Results showed no relationship between repeated-

extinctions and total keypecks during extinction after CRF (r--.08), VR reinforcement (r=-

.20) or the extinction ratio (r=-.18).






Figure 7 shows the number of keypecks emitted during the extinction component

previously paired with VR reinforcement (upper left panel) and CRF (upper right panel) as

a function of the number of reinforcers per component presented during VR CRF. The

number of keypecks emitted during the extinction component previously paired with VR

reinforcement as a function of the number of reinforcers per component presented during

both the VR VR and VR CRF phases is shown in the lower left panel of Figure 7. These

analyses show both between- and within-subject variation. The number of responses

emitted during the extinction component previously paired with CRF was highly variable.

Pigeons 3180 and 3519 showed no systematic effect due to the number of previous

reinforcers delivered during VR CRF, but Pigeons 2842 and 2875 showed decreases as the

number of previous reinforcers during VR CRF increased. The number of responses

emitted after VR reinforcement was also highly variable and showed no systematic

relationship (r = -.03) to number of reinforcers per component presented during VR CRF,

nor to the number of reinforcers per component presented during VR VR and VR CRF

combined (r = .22). The extinction series for Pigeons 2374, 2822 and 2880 were only

successfully replicated once, making interpretations of trends for their individual data

impractical.

Figure 8 shows the extinction ratio as a function of the number of reinforcers per

component presented during the VR CRF phase. In the present experiment the extinction

ratio generally increased as the number of reinforcers during VR CRF increased (b = .71, r

= .83). The dotted lines in Figure 7 represent the 1.00 value of the extinction ratio (i.e., no

PREE). The dashed lines of Figure 7 represent the extinction ratio for the extinction series

in which there was no VR CRF phase (3180-6 and 3519-7).

Figure 9 shows the extinction ratio determined by six different extinction criteria as

a function of the number of reinforcers presented during VR CRF. Each criterion required

a continuous period of time without a response (i.e., 5 sessions, 1 session, or 1

component). In addition, these criteria could be "linked" to both components of the













gS o S u
uo, -g




"o .9 o


$ O o




040
0 4.) 0 t






r c) P> cn > ''






, 0 ) 0

aC .5 0 ) 4


4 ) 8 = 54
(A 0 D S < 0 4.)
0 c i0 'i r










oa i t
4 0










- 4) ;- 0
S2 >o- 2 -g

o c> U H






S5 0 u E H-



LL*a 0o o 0 0 o
0 U=
.~ r, nu
,~i~ e E E
1E OtC00 0










































00
s) 00












4 o _______U

























sioad~av OOC






Figure 2. Cumulative extinction curves for Pigeons 2822, 2374, and 2875. The ordinates
show the total number of responses emitted during the extinction component previously
paired with VR (solid) and CRF (dotted). Note that the ordinates change for each pigeon.
The abscissa shows consecutive daily sessions from the beginning of the extinction phase
(EXT EXT) of each extinction series. Each panel number corresponds to the order in
which the extinction series were presented for each pigeon. Below the panel number is
the number of reinforcers per component delivered during VR CRF. The "x" is the value
of the VR schedule used during the VR VR and VR CRF phases of the individual
extinction series (i.e., x = 100 for Pigeons 2822 and 2374 for all extinction series; for
Pigeon 2875, x = 100 for the first extinction series and x = 25 for the second and third
extinction series).











2822 (x=100)

1 -- VR x
580 ...... CRF


237

1
3250




! !


- I I


I *I I I 2 I
4 (x=100)

2
1550


2875

1 2
580 2500
(x=100) (x=25)




0 20 40 60 80 100


(x=25)


, I I


I I I


0 20 40 60


80 100
Sessions


O

0
c-
--


0
h-


U)
Y

-Q)

O
0
>4-




0
F-


S | I I I






Figure 3. Cumulative extinction curves for Pigeons 2842 and 2880. The ordinates show
the total number of responses emitted during the extinction component previously paired
with VR 50 (solid) and CRF (dotted). The abscissa show consecutive daily sessions from
the beginning of the extinction phase of each extinction series. Each panel number
corresponds to the order in which the extinction series were presented for each pigeon.
Below the panel number is the number of reinforcers per component delivered during VR
CRF. The asterisk "*" in panel 2 for Pigeon 2842 designates the termination of the
extinction phase before reaching the extinction criterion (i.e., 5 consecutive sessions
without a response in either component). The plus sign "+" in panel 1 for Pigeon 2880
designates an apparatus failure (see text) during VR CRF of the extinction series.










2842


1
900


3
3000


-- VR50
........ C R F


2*
2020


I I

4
40


I ,


2880
1+8
1820


3
1260


2
220


0 20 40 60 80


0 20 40 60 80
Sessions


O
cn



0
-c-


' '


I'


| i














3180


VR100
- -........ CRF





I I i I


1
2250





I I


(I)
nU)
-o
c-

0
0



c-
C-




(.



4-d
>0
0)



0
0
t-


10

5

0
20

15

10

5

0
20

15

10

5

0


I I. I


0 10 20 30 40 50 60 70


2
575


4
1225


7 I I I I I


0 10 20 30 40 50 60 70


Sessions




Figure 4. Cumulative extinction curves for Pigeon 3180. The ordinates show the total
number of responses emitted during the extinction component previously paired with VR
100 (solid) and CRF (dotted). The abscissa show consecutive daily sessions from the
beginning of the extinction phase of each extinction series. Each panel number
corresponds to the order in which the extinction series were presented. Below the panel
number is the number of reinforcers per component delivered during VR CRF. The
asterisk "*" in panel 3 designates the termination of the extinction phase before reaching the
extinction criterion.


5
25

.. .. .


, I .


, I ,


I I I111111111






Figure 5. Cumulative extinction curves for Pigeon 3519. The ordinates express the total
number of responses emitted during the extinction component previously paired with VR
50 (solid) and CRF (dotted). The abscissa expresses consecutive daily sessions from the
beginning of the extinction phase of each extinction series. Each panel number
corresponds to the order in which the extinction series were presented. Below the panel
number is the number of reinforcers per component delivered during VR CRF. The
asterisks "*" in panels 2, 3, 4 and 6 designate the termination of the extinction phase
before reaching the extinction criterion. The plus sign "+" in panel 4 designates an
apparatus failure during VR CRF of the extinction series.












VR50 2*
.......- CRF 300


4*+
2250


S I .


6*
25





i' I i


0 10 20 30 40 50 60 70


0 10 20 30 40 50 60 70


Sessions


3519


1
1250


U)
-0

0
"-
0
[--'
-4-


3*
3225


I I I


5
1225





- i- i i-i i i- i i-i i i















Continuous Reinforcement


15000


10000


5000


0


Variable Ratio Reinforcement


15000


10000


5000


0


v b=-0.09
r=-0.18

0*
-0

;I -V-- .......
I I I I I I I
1 2 3 4567

Extinction Series


o 2374
* 2822
v 2842
* 2875
A 2880
a 3180
v 3519


Figure 6. Total keypecks in extinction after CRF (upper panel) and total keypecks in
extinction after VR reinforcement (middle panel) as a function of the presentation order of
the extinction series. The lower panel shows the extinction ratio as a function of the
presentation order of the extinction series. Each panel shows the data points for all
extinction series except those not meeting the extinction criterion as well as those following
an apparatus failure during VR CRF. "r" is Pearson's r and "b" is the slope of the
regression line (method of least squares).


b=-129.51
r=-O.OB
*

-I *
a vv
TV
I V I I I I


0 b=-400.42
r=-0.20



-0O
- a
v v
0 M V
o -
I I I I I I














CRF Component


0 1 2 3


b=-0.89
r=-0.31
0

S v

I 1 r I
0 1 2 3


Reinforcers per Component During VR CRF
(in thousands)


0123456789


0 2374
* 2822
v 2842
* 2875
A 2880
D 3180
v 3519


Reinforcers per Component
During VR VR and VR CRF

(in thousands)







Figure 7. Total keypecks in extinction after VR reinforcement (upper left panel) and total
responses during extinction after CRF (upper right panel) for each extinction series as a
function of the number of reinforcers per component during VR CRF. The lower panel
shows the total keypecks in extinction after VR reinforcement for each extinction series as a
function of the number of reinforcers per component during both the VR VR and VR CRF
phases. Each panel shows the data points for all extinction series except those not meeting
the extinction criterion as well as those following an apparatus failure during VR CRF. "r"
is Pearson's r and "b" is the slope of the regression line.


15000


10000


5000

0


n b=-0.11
r=-0.03
D D



S0
I I I I I


15000


10000


5000


b=0.42
* r=0.22
Sa 0


J 0A 0

I I I -I -I -I -I I


VR Component






Figure 8. Total responses during extinction expressed as the extinction ratio ordinatee) for
each extinction series as a function of the number of reinforcers per component during
VR CRF abscissaa) for each pigeon (2374, 3180, 3519, 2822, 2842, 2875 and 2880). The
dotted line at 1 on the ordinate designates no PREE. Values greater than 1 indicate a
PREE. Values less than 1 indicate a reversed PREE. The dashed line for Pigeons 3519
and 3180 highlight the extinction ratio after exposure to the VR VR phase only. The
asterisk "*" designates the termination of the extinction phase before reaching the
extinction criterion. The plus sign "+" designates an apparatus failure during the VR
CRF phase of the extinction series. The panel labeled "ALL" shows the data points from
the individual plots except those not meeting the extinction criterion as well as those
following the apparatus failure during VR CRF. "r" is Pearson's r and "b" is the slope of
the regression.
















2374 o



O0
2






3519

3.
76. V
56*


.2+ .
7





2842






4
I I I----------------
t4---- ] ***,-- I I


2822




....e.................................
21





2875



2



3 I


5
2880 ALL r=0.83
S4 b=0.71
1+
3

S2 -

................................
I I I 0 -

0 1 2 3 0 1 2 3

Reinforcers per Component During VR CRF


3180

S3*0
1
6 02
.........................






Figure 9. Total keypecks during extinction expressed as the extinction ratio ordinatee) for
each extinction series as a function of the number of reinforcers per component during
VR CRF abscissaa) for each extinction criterion. "5 sessions" equals 5 consecutive
sessions without a keypeck, "1 session" equals 1 session without a keypeck, "1
component" equals I component without a keypeck. "Linked" refers to the requirement
that the time-without-a-keypeck criterion occurred for both extinction components.
"Unlinked" refers to the independent assessment of the extinction criterion between the
components. The solid line at 1 on the ordinate designates no PREE. "r" is Pearson's r,
"b" is the slope of the regression line and "n" equals the number of extinction series that
met the extinction criterion.













Linked Unlinked
5 Sessions

Sn=20 n=21
b=0.71 b=0.68
r=0.83 r=0.80
L.0

1 1
ry


r 0.1
o
1 Session
cJ
C
-- 10 n=25 _n=25 o 2374
x b=0.608 b=0.604
j r=0.70 r=0.67 2822
c U U.^-^ T2842
1n 9 2875
0 A 2880
a, 3180
0.
> v 3519
, 0.1 I
1 Component
o
10 n=25 "n=25
b=0.90 b=0.456
0 r=0.49 r=0.12
0 0 0
3 1 '- -



0.1 I I
0 1 2 3 0 1 2 3

Reinforcers per Component During VR CRF


(in thousands)





34

multiple schedule. For example, a linked criterion would stipulate that a response could not

occur for 5 consecutive sessions in either component of the multiple schedule. An unlinked

criterion would be the termination of the extinction phase for each component separately.

Thus, once an extinction criterion was met in the first component of a two component

multiple schedule, responses would no longer be counted for the first component, but

responses for the second component would continue to be counted until the extinction

criterion was met for the second component. Figure 9 shows that the more stringent the

extinction criterion (i.e., longer time without a response and linked to both components) the

stronger the correlation between the extinction ratio and the number of reinforcers delivered

during VR CRF.

Extinction performances were also measured in terms of the changes in response

rate as the log proportions of the response rates of the first extinction session (Nevin,

Smith, & Roberts, 1987). Response rates generated by VR reinforcement were usually

higher than response rates produced by CRF. Proportion of initial extinction responding

was used to adjust for these initial differences produced by the different schedules of

reinforcement. The slopes resulting from a regression analysis (method of least squares)

were used to calculate the rate of change from the baseline conditions. The smaller the

value of the slope (b) the greater the behavioral momentum. This relationship between the

slope measure and behavioral momentum required the extinction ratio for the slopes to be

inverted (CRF/VR) to make it comparable to the extinction ratio of the number of extinction

responses (see above). A slope extinction ratio of 1.00 indicates no PREE. Values greater

than 1.00 indicate a PREE and values less than 1.00 indicate a reversed PREE.

Figure 10 shows the linear regression lines fitted to the log proportion data for two

pigeons. Pigeon 2842, series 4, and Pigeon 3519, series 1, are examples of the reversed

PREE and the PREE, respectively. Included in each panel is the value of the slope

extinction ratio.






Figure 11 shows the slope extinction ratio for log proportions as a function of the

number of reinforcers delivered during VR CRF. Data from all the pigeons were included

except for Pigeons 2880 series 1 and 3519 series 4, due to an apparatus failure during VR

CRF (see method). The ratios of the slopes of the log proportions were uncorrelated with

the number of reinforcers per component delivered during VR CRF (r = -.01), but the

variability in the slope extinction ratios seemed to decrease as the number of reinforcers per

component during VR CRF increased.

Figure 12 shows an estimate of the "savings" in the number of keypecks emitted

during extinction by implementing CRF after VR reinforcement before proceeding with

EXT. The measurement for this analysis was derived by adding the number of keypecks

emitted during extinction after the CRF history plus the number of keypecks emitted during

the CRF component of the VR CRF phase (CRF adjusted) and then subtracting the number

of keypecks emitted during extinction after the VR history (CRF adjusted VR). This

measure takes into account the number of keypecks required during CRF training to

produce the absolute reduction during EXT. The measure is still biased, however, in

showing a "savings" because additional VR reinforcement occurred while the pigeon was

exposed to CRF. A zero score indicates no effect of the CRF schedule in increasing or

decreasing the number of keypecks during extinction compared to extinction after VR

reinforcement. Values greater than zero indicate an increase in the number of keypecks

during extinction after CRF compared to extinction after VR reinforcement. Values less

than zero indicate a decrease in the number of keypecks during extinction after CRF

compared to extinction after VR reinforcement. The results showed a variable effect.

Extinction performances after exposure to a small number of CRF reinforcers (0 to 40

reinforcers) showed a reduction in keypecking during EXT after CRF by as much as 1516

keypecks, but also produced increases in keypecking as great as 1390 keypecks. Total

keypecks after CRF were consistently fewer (i.e., by 127 to 4320) when CRF lasted

between 90 to 1550 reinforcers. Only one of the four extinction series in which the number




36

of CRF reinforcers was greater than 1550 showed a reduction in keypecking during EXT

when keypecks during the CRF component of the VR CRF phase were included.
















2842-4
0.45

U,

i_ 0 0 .1" s-



1.44







~ \0.1 -
0 .0
r o
0.E .01
-- VR
x
.,.., ..- -. C R F
T 3519-1
I5 1.44

O ""

s ar VR. O d s .
0b-

0 .0 1 ,' .
1 5 10

Sessions













Figure 10. Individual extinction performances (2842-4 and 3519-1) expressed as the log
proportion of keypecks per minute during the first extinction session across the first 10
sessions of the extinction phase abscissaa). Closed circles represent extinction
performances after VR. Opened circles represent extinction performances after CRF.
Regression lines were fitted by the method of least squares. Solid lines represent the slope
of the extinction performance after VR and dotted lines represent the slope of the extinction
performance after CRF. The numbers in the upper right of each panel are the value of the
slope extinction ratio (CRF/VR).






Figure 11. Rate of keypecking during extinction expressed as the slope extinction ratio
(CRF/VR) of the log proportion as a function of the number of reinforcers per component
during VR CRF. The dashed line for Pigeons 3519 and 3180 highlight the slope
extinction ratio after exposure to the VR VR phase only. The dotted line at 1 on the
ordinate indicates no difference in the slopes. Values greater than 1 indicate a PREE.
Values less than 1 indicate a reversed PREE. A regression line (method of least squares)
was fitted the slope extinction ratio of the log proportions (solid line).

















2374




0
2 0
1

fI I



3519
v
2*


'F5
--------------------4w,------V
7 3*





2842





v V
1 2* 3
I I



2880




A A
2 3
1+


0 1 2 3


3180




2
S5 0 3.o

I I
[ I- ---- -----* .---------- I--- ---- -- -





2822






2 *





2875






--- g ---- a
I312



ALL r=-O.01
b=-0.004



S
0
a
I I

0 1 2 3


Reinforcers per component During VR CRF


(in thousands)














2000


1000 0 o





-1000 A


-2000


-3000- 0 r


-4000 -


-5000
0 1000 2000 3000


Reinforcers per Component During VR CRF














Figure 12. Total keypecks during extinction expressed at the difference between keypecks
after VR reinforcement and keypecks after CRF plus keypecks during the CRF component
of the VR CRF phase (i.e., CRF adjusted VR) as a function of the number of reinforcers
per component during VR CRF. The dotted line at zero on the ordinate designates no
difference between the two components during extinction. Values greater than zero indicate
more absolute keypecks during the VR history component compared to the CRF history
component. Values less than zero indicate less absolute keypecks.










DISCUSSION

The PREE, as measured by the total number of responses during extinction, was

replicated both within and between subjects using a single-subject, free-operant

experimental design. These results conflict with previous attempts at generating a free-

operant within-subject PREE that produced reversed PREEs (Adams, Nemeth, & Pavlik,

1982; Pavlik & Carlton, 1965; Pavlik, Carlton, Lehr, & Henrickson, 1967). The

difference between the present and previous attempts at producing a free operant within-

subject PREE may have been due to the extensive amount of exposure to the schedules of

reinforcement in the present study as well as the equating of the number of reinforcers

delivered during both the CRF and PRF schedules.

Equating the number of reinforcers in between-subject experiments has not been

necessary to produce the PREE (see Jenkins & Stanley, 1950, for review). The extent to

which the PREE can be produced using a within-subject procedure without equating the

number of reinforcers presented during the PRF and CRF schedules has not been explicitly

examined.

Previous research exposing subjects to different amounts of reinforced training

(i.e., number of reinforcers) has shown that responding during extinction persists longer

the greater the number of reinforcer presentations (CRF, Perin, 1942; Williams, 1938, and

PRF, Hothersall, 1966). Pavlik, Carlton and Manto (1965) compared a within-subject

procedure that equated the number of responses (60) to a procedure that equated the

number of reinforcers (40). Both procedures produced a PREE, but the procedure

equating the number of reinforcers produced greater total extinction responses after both the

CRF and VR 3 schedules as well as producing a larger PREE as estimated by the difference

in the total extinction responses. Although it may not be necessary to equate the number of






reinforcers between a CRF and PRF schedule in order to produce a within-subject PREE,

the practice of equating the number of reinforcers may accentuate the PREE.

The production of the PREE with a single-subject design shows that the analysis of

the phenomenon of the PREE is appropriate at the level of individual subjects. Although

the PREE was consistently produced, several issues are raised when a single-subject

design is used to investigate a behavioral phenomenon. The first of these issues deals with

the repeatability of the phenomenon in a single subject. Previous research showed that

repeated exposure to CRF (40 reinforcers) and 60 or 90 minutes of extinction reduced lever

pressing during extinction when both training and extinction occurred on the same day

(Bullock & Smith, 1953; Clark & Taylor, 1960). Bullock (1960) showed that intermittent

schedules (20 reinforcers of FR10 or Fixed interval 26 s) also showed the repeated-

extinction effect. Wickens and Miles (1954) exposed rats to training (15 reinforcements)

and extinction (1 hour) on alternate days and reported an initial increase in bar pressing

during the second exposure to extinction followed by a slow steady decrease during

successive replications of extinction. Anger and Anger (1976) extended the training phase

to two days (60 reinforcements) and the extinction phase to 8 days (100 trials per day) and

reported no reduction in total number of keypecks during extinction with successive

extinction series replications. Anger and Anger (1976) did see a repeated-extinction effect

when they limited the measurement of keypecking to the first extinction session. The

repeated-extinction effect may be a phenomenon limited to the early part of extinction that is

not seen when data are summed over a long extinction exposure (e.g., 8 daily sessions).

In the present experiment, total keypecks after CRF and VR and the extinction ratio

showed no consistent repeated-exposure effect when keypecking was carried to the

extinction criterion (i.e., 5 consecutive session without a response). In an analogous

comparison of extinction performance during the early part of the extinction phase (Anger

& Anger, 1976), total keypecks were counted for the first extinction session. The results

were the same as when keypecking was counted over the entire extinction phase.





43

The differences in the effects of repeated-extinction between previous research and

the present experiment may be due in part to the extensive (e.g., months of daily sessions)

exposure to reinforcement during the VR VR and VR CRF phases before the

reimplimentation of a fairly long exposure to extinction. One result of particular interest is

that there was more of a relationship between repeated-extinction and keypecks during the

VR component than there was during the CRF component. Changing of the reinforcement

schedules between the VR VR and VR CRF phases may have acted to disrupt the stimulus

control that could be produced by the simple alternation between a single reinforcement

schedule and extinction.

The amount and type of retraining necessary to minimize the effect of the previous

exposure to extinction was not explicitly examined in the present experiment. The results,

however, suggest that the extinction of behavior may be a reversible effect if the organism

is exposed to multiple reinforcement schedules in the presence of the same discriminative

stimulus for an extended period of time between exposures to extinction.

Previous free-operant research exposing subjects to either CRF (Margulies, 1961;

Hearst, 1961; Perin, 1942; Williams, 1938) or PRF (Hothersall, 1966; Wilson, 1964)

showed that as the number of reinforcers during acquisition increased, the number of

responses during extinction increased. In the present study, the extinction ratio showed a

strong increasing trend as the number of reinforcers per component during VR CRF

increased. Increases in the total number of keypecks emitted during extinction was not

replicated, however. The number of responses after CRF showed a weak decreasing trend

as the number of CRF reinforcers during VR CRF increased. The number of responses

emitted during extinction after VR reinforcement showed no relationship to the number of

reinforcers presented by a VR schedule during the VR CRF phase or to the number of

reinforcers per component in both the VR VR and VR CRF phases combined. The number

of reinforcer presentations in previous experiments were small in comparison to the present






experiment and a lack of a replication may be due in part to the larger parameter values

examined in the present experiment.

The lack of consistency between the total keypecks measure of the present

experiment and the number of responses or trials in previous experiments may have been

due to the use of the multiple schedule in the present experiment. Exposing individual

subjects to both VR and CRF histories of reinforcement may have produced an interaction

between the two schedules of reinforcement (e.g., behavioral contrast or induction). To

help reduce the interactions between the components, the keylight and houselight were

darkened (i.e., 1 min. time-outs) between the components. The darkened keylight was

depressed on occasion (e.g., about once or twice every ten sessions), but the lack of

continued and persistent responding during the one minute time-out interval reduces the

possibility of the development of chained responding between the VR and CRF

components during the VR CRF phase.

Interactions could also have been detected when changing from VR VR to VR

CRF. Changing one component from VR to CRF consistently decreased the "rate" of

responding, more accurately described as average reciprocal latency, for the CRF

component. The effect on the terminal performance of the VR component (e.g., five days

before the initiation of the EXT EXT phase) was varied (see Appendix B). Some pigeons

showed consistent increases in keypecking rates during the VR component when CRF was

scheduled in the alternate component (2374 and 2822), or consistent decreases (2880), but

most of the pigeons showed varied effects (3519, 2875, 3180, and 2842). In view of

these results, it seems unlikely that the lack of an effect of the number of reinforcers per

component during VR CRF on the number of keypecks during extinction was due to an

interaction between the components.

The addition of an extensive training regimen (i.e., VR VR) before the presentation

of the different schedules of reinforcement may account for the differences between the

present and previous experiments that did not incorporate extensive pretraining. Hothersall






(1966) assessed the role of CRF pretraining on extinction responding after maintaining

behavior on a VR 4 schedule of reinforcement and found no difference between groups

exposed directly to a VR 4 schedule after response shaping (i.e., no pretraining) and

pretraining with 10 CRF reinforcers before exposure to a VR 4 schedule. Comparing

groups receiving 10, 50 or 100 pretraining CRF reinforcers showed that the 50 and 100

CRF groups emitted fewer responses during extinction than the group receiving only 10

CRF reinforcers. Groups receiving 50 or 100 CRF reinforcers showed no differences

during extinction. This difference in total responses between the group receiving 10 and

groups receiving 50 or 100 CRF reinforcers was evident even after 200 reinforcers were

delivered during VR 4 schedule. Hothersall's results showed that a sufficient amount of

exposure to a previous history of reinforcement could have a pronounced affect upon

extinction responding.

The present procedure exposed each pigeon to an extensive history (see Table 2) of

VR reinforcement as pretraining during the first phase of each series. If it is assumed that

the relationship between the number of PRF reinforcers and the number of responses

emitted during extinction is an increasing, negatively accelerating function that approaches

an asymptote, then the lack of an increase in the number of extinction responses after VR

reinforcement could be attributed to the large number of reinforcers delivered during the VR

schedule during both the VR VR and VR CRF phases. Examination of the number of VR

reinforcers presented during the component paired only with VR reinforcement showed a

highly variable but flat function in relation to the number of keypecks during extinction for

the component previously paired with VR reinforcement. These results suggest that the

range of the values for the number of reinforcers presented under a VR schedule (200 to

8750) were for the most part above the point where additional reinforcement presentations

would have produced additional responses during extinction.

The decrease in extinction responses after the CRF history as a function of the

number of reinforcers per component during the CRF VR phase could also be explained by






the pretraining history of VR reinforcement. If CRF produces less extinction responding

than VR reinforcement, then implementing a CRF schedule between a VR schedule and

extinction should decrease the total number of responses emitted during extinction. This

effect was seen in the total number of keypecks during extinction measure, and was also

represented in the magnitude measure of the PREE (i.e., the extinction ratio).

The interpretation from between-subject procedures examining the PREE frequently

has been based upon an averaged sample of performances by groups of subjects (e.g.,

Adams, Nemeth & Pavlik, 1982; Amsel, Rashotte & MacKinnon, 1966). The group

averaging of individual data was done, in part, to highlight the effects of the independent

variables that might be hidden by the variability produced by individual differences.

"Individual differences" is still a relevant issue with the single-subject design concerning

the within-subject replication of the PREE. Each extinction series was preceded by a

unique history (e.g., none or a number of previous extinction series). One way to

highlight the effects of potentially relevant independent variables that might be hidden by

differences between individuals and replications is to use the extinction ratio as a measure

of the magnitude of the PREE. The potential utility of the extinction ratio is substantiated

by its sensitivity to the number of previous reinforcers per component during VR CRF.

Whether the extinction ratio continues to be a useful dependent variable will depend upon

its sensitivity to other parametric manipulations.

The slopes of the log proportion were highly variable from condition to condition

and did not change systematically as a function of the number of previous reinforcers

delivered during VR CRF. The variability of the log proportion measures appeared to

decrease as the number of reinforcers per component during VR CRF increased, but this

observation may be the result of fewer replications of the extinction series at large values of

the number of reinforcers during VR CRF.

Several extinction criteria for the present experiment were based upon arbitrary but

somewhat natural lines of fracture (i.e., 1 component, 1 session, or 5 sessions). As the






criterion became more lenient (i.e., less continuous time without a response), increased

variability was observed in the magnitude measure of the PREE (i.e., the extinction ratio).

Independent (i.e., unlinked) determination of "extinction" between the components also

increased the variability of the magnitude measure. These results suggest that part of the

variability reported by Williams (1938), Perin (1942) and others may have been due to the

premature termination of the extinction procedure. Until there is a more thorough analysis

of the role that extinction criteria play in the production of the PREE, criteria based upon

time without a response should be fairly conservative (e.g., 3 or 4 successive 20-min

sessions) and be tied to both CRF and PRF extinction performances to equate the amount

of exposure to extinction.

If interpolated CRF is effective in reducing the number of responses emitted during

extinction after PRF, the implementation of CRF before EXT could be a useful tool in the

reduction of behavior in applied settings. The results from the present experiment suggest

that the use of CRF prior to EXT when behavior has been previously maintained on an

intermittent schedule of reinforcement can reduce the number of responses emitted during

extinction. Adjusting for the number of responses emitted during CRF showed that

reductions in keypecking during extinction were consistently produced when at least 40 but

fewer than 1500 CRF reinforcers preceded the extinction procedure.

A limitation of the application of interpolated CRF to the reduction of behavior via

extinction is that the reinforcer that is maintaining the target behavior is not always known.

A few experimenters have attempted to circumvent this problem by using CRF with a

known reinforcer before implementing extinction (Foxx & McMorrow, 1983; Neisworth,

Hunt, Gallop & Madle, 1985; Schmid, 1986). Although responding after the removal of

the known reinforcer was reduced, the effect was usually temporary. The brevity of the

effect was most likely due to the continued reinforcement by the intermittent schedule that

was maintaining the behavior before the introduction of the known reinforcer (Wylie &

Grossmann, 1988).






The present experiment illustrates a procedure for the repeated production of the

PREE within a single organism. It seems safe to conclude from the results of the present

experiment that the PREE is a behavioral process (Sidman, 1960) existing at the level of

individual organisms. The present procedure can therefore be used to assess a range of

parameters of a variety of independent variables on a continuous measure of the magnitude

of the PREE (i.e., the extinction ratio). For example, the present experimental procedure

could be used to assess more accurately the effects of repeated exposures to extinction. If

the repeated-extinction effect is minimal (i.e., the extinction ratio remains unchanged) a

baseline would then exist to test a range of stimulus values (e.g., drug dosage) on behavior

undergoing extinction using a single organism.

The emphasis of future single-subject research on the PREE should begin by

isolating further the conditions necessary for the consistent production of the PREE within

the single organism. Parametric analysis of the variables that might contribute to the

magnitude of the PREE (e.g., magnitude of reward, deprivation, different kinds of

intermittent schedules) will then provide the necessary foundation on which to build the

prediction, control, and interpretation necessary for a more complete understanding of the

PREE.













APPENDIX A
COMPONENT DURATIONS (MIN) DURING EXTINCTION







3519 VR CRF 3180 VR CRF 2374 VR CRF 2880 VR CRF

1 1.84 2.01 1 4.35 3.84 1 4.83 3.82 1 2.90 3.73


1.72
1.79
1.96
1.84


2.01
2.45
2.34
2.87


4.47
5.59
6.43
6.45


5.13
5.70
6.03
6.92


11.68 total 27.29 27.62


total 9.15

2 2.14
1.81
2.09
1.71
1.59

total 9.34

3 1.82
1.57
1.47
1.23
1.18

total 7.27

4 1.74
total 8.70

5 1.69
total 8.45

6 1.55
total 7.75

7 1.59
total 7.95


2 4.33
4.25
4.17
4.79
4.99


3.62
4.00
5.02
5.21
4.78


total 22.53 22.63


5.50
6.82
5.53
5.67


5.99
6.03
5.88
5.87


total 28.35 27.59


2 3.56
3.50
3.72
3.94
4.13


3.49
3.80
4.67
4.70
4.91


total 18.85 21.57


3 7.57 7.57
total 37.85 37.85 2875 VR CRF


4 5.86 5.86
total 29.30 29.30


2.04
1.68
2.20
1.45
1.91

9.28

1.60
1.86
1.92
1.79
1.58

8.75

1.74
8.70

1.69
8.45

1.55
7.75

1.59
7.95


6.43
32.15


6 6.49 6.49
total 32.45 32.45


1 4.66 4.67
4.83 5.54
6.38 6.74
6.39 6.61

total 22.26 23.56


2 1.05
total 5.25

3 1.11
total 5.55


1.40
7.00

1.11
5.55


total 14.50 18.65

2 3.25 3.25
total 16.25 16.25

3 3.66 3.66
total 18.30 18.30


2842 VR CRF

1 2.98 2.98
total 14.90 14.90

2 2.80 2.80
total 14.00 14.00


3 3.02
total 15.10

4 2.49
total 12.45


3.02
15.10

2.49
12.45


2822 VR CRF

1 6.27 6.73
total 31.35 33.65

2 5.35 5.35
total 26.75 26.75


5 6.43
total 32.15













APPENDIX B
STABILITY AND VARIABILITY STATISTICS OF THE
VR VR AND VR CRF PHASES.












4.)

S o 00 -

00*C- 000- -f 0
N- cc 00 *O -00 Cfrr
Is !"0 io"
I
,*a
a ggg ""g -4 s '
oo


S o ( r
o(00 N~ N



~cei

a BgSS SSS 88 S8






N s y
as oa s-sa
0 0
x -t -o in/ oi -o <0 (1 .)

sa- -- o^ -o ~"CI- d o--o 'n


--- -" (










m O'
o (1000 N000 Cr0 0000 4 -0 -'
a
50 0" C~ 00 'C ^ /^'



S r 00 -
0 0 0












00000 44 0000 000( 4. )4444 AS
3 r00 OC.0 r^- 00 : -3 M O
-- >- --O '-- r--- p
|500 00 r- C C

1 0
= ?? ES Cg00 (5N 00000 Crg CO











o -= 0-00 0000 000 4 g o -= 4 -
.04-_ 00 -< -- 0-- X
4.)

C r-, -" (o0














mE 255 88 B 88 8 CER


o
S N 0 S 0 N 0 00
So -' (1 -' f( -' 0000




I U -

0- -0 0-- v 4
0
00 C- 00 a 0 -c C (1 CT 00 0 4. n r
0 -: ri -; 0 -- S J
000 0-















,,t
S; ~ C8 N0 NS N8 0 CC u
---------0- 00 O.**' O.** '-.0 ~ -

00 00 0 C1 N9 N00 N 0
r- -
*Sggo- 0 -i r o- d- d -0 No C --o*-_ o



U~ 0 00c r o
g 0 (1 -- 00 ^ (0 r- (1 0- 00 C l
N00 (100C 001 (00 Cl0 C 0 C1 000
g. 3~ GS 8 8 s 8 ~a9"BP '.e
'0 00 00 000 00 0000.o r00---o -00 00r~ 0loo '00^' 000f rZZ 0 r~im
aC o'.=- -< t-0' 006 C-i~ C-0- C--CT 0 C- -0 -*'oo' .- 0- -o 4) o
2 L */>-o V-^-o 5--d -; --: o<--^ OQ -;-"; o-^ -;c: d; --o ~ <;- w o o*3
-wl oo ~ o e. -^ C C 4




c E
-6 nr
^ 0 ------------o-5 N-- 00-0 0 pi 0
Bri O n00 0000 'dCT Cno 00 00 C-0 NO O 0* 00 N- N N :T ^ (.
a ( o m 0 O o-- *0>-" 0 0 ;* ." 4.) /





CC CU CC CCEza U 1
C C C CC Cc CC CC CCi CCr ^ < r- o f

( C 003 WOO0 OOC01 O~01 l-OO CT- r-C00 COO- --.=7 WO I** O~ ^1 =- "


('a a C c C N c
00 O P^0 OBU O)U OXO Ui< itU OYU OXU O U 5






)N n n t- 0
M Z^














REFERENCES


Adams, J. F., Nemeth, R. V., & Pavlik, W. B. (1982). Between- and within-subjects
PRE with sucrose incentives. Bulletin of the Psvchonomic Society, 20, 261-262.

Amsel, A., Rashotte, M. E., & MacKinnon, J. R. (1966). Partial reinforcement effects
within subject and between subjects. Psychological Monographs: General and
Applied, 628, 1-39.

Anger, D., & Anger, K. (1976). Behavioral changes during repeated eight-day
extinctions. Journal of the Experimental Analysis of Behavior, 26, 181.

Bullock, D. H. (1960). Repeated conditioning-extinction sessions as a function of the
reinforcement schedule. Journal of the Experimental Analysis of Behavior, 3, 241-
243.

Bullock, D. H., & Smith, W. C. (1953). An effect of repeated conditioning-extinction
upon operant strength. Journal of Experimental Psychology, 46, 349-352.

Clark, F. C. (1964). Effects of repeated VI reinforcement and extinction upon operant
behavior. Psychological Reports, 15, 943-955.

Clark, F. C., & Taylor, B. W. (1960). Effects of repeated extinction of an operant on
characteristics of extinction curves. Psychological Reports, 6, 226.

Dyal J. A., & Sytsma D. (1976). Relative persistence as a function of order of
reinforcement schedules. Journal of Experimental Psychology: Animal Behavior
Processes, 2, 370-375.

Feider, A. (1973). Within subjects partial reinforcement extinction effects for a bar
pressing task. Canadian Journal of Psychology, 27, 356-366.

Foxx, R. M., & McMorrow, M. J. (1983). The effects of continuous and fixed ratio
schedules of external consequences on the performance and extinction of human
stereotyped behavior. Behaviour Analysis Letters, 3, 371-379.

Gellermann, L. W. (1933). Chance orders of alternating stimuli in visual discrimination
experiments. Journal of Genetic Psychology, 42, 206-208.

Hearst, E. (1961). Resistance-to-extinction functions in the single organism. Journal of
the Experimental Analysis of Behavior, 4, 133.

Hothersall, D. (1966). Resistance to extinction when continuous reinforcement is
followed by partial reinforcement. Journal of Experimental Psychology, 72, 109-
112.






Iversen, I. H., & Lattal, K. A. (Eds.) (1991). Experimental analysis of behavior:
Techniques in the behavioral and neural sciences, Vol. 6. Elsevier: New York.

Jenkins, H. M. (1962). Resistance to extinction when partial reinforcement is followed by
regular reinforcement. Journal of Experimental Psychology, 64, 441-450.

Jenkins, H. M., & Stanley, J. C. (1950). Partial reinforcement: A review and critique.
Psychological Bulletin, 47, 193-234.

Johnston, J. M., & Pennypacker, H. S. (1980). Strategies and tactics of human
behavioral research. Hillsdale, NJ: Lawrence Erlbaum Associates.

Keller, F. S. (1940). The effect of sequence of continuous and periodic reinforcement
upon the "reflex reserve." Journal of Experimental Psychology, 27, 559-565.

Kimble, G. A. (1961). Conditioning and learning. New York: Appleton-Century-Crofts,
Inc.

Margulies, S. (1961). Response duration in operant level, regular reinforcement, and
extinction. Journal of the Experimental Analysis of Behavior. 4, 317.

Moreland, J. W., Stalling, R. B., & Walker, L. C. (1983). The effect of interpolated
continuous reinforcement on persistence in extinction: A replication demonstrating
reversibility. Behavior Analysis Letters, 3, 149-156.

Neisworth, J. T., Hunt, F. M., Gallop, H. R., & Madle, R. A. (1985). Reinforcer
Displacement: A preliminary study of the clinical application of the CRF/EXT
effect. Behavior Modification. 9, 103-115.

Nevin, J. A. (1974). Response strength in multiple schedules. Journal of the
Experimental Analysis of Behavior. 21, 389.

Nevin, J. A., Mandell, C., & Atak, J. R. (1983). The analysis of behavioral momentum.
Journal of the Experimental Analysis of Behavior. 39, 49-59.

Nevin, J. A., Smith, L. D., & Roberts, J. (1987). Does contingent reinforcement
strengthen operant behavior? Journal of the Experimental Analysis of Behavior,
48, 17-33.

Pavlik, W. B., & Carlton, P. L. (1965). A reversed partial-reinforcement effect. Journal
Experimental Psychologv, 70, 417-23.

Pavlik, W. B., Carlton, P. L., Lehr, R., & Hendrickson, C. (1967). A reversed PRE.
Journal of Experimental Psychology, 75, 274-276.

Pavlik, W. B., Carlton, P. L., & Manto P. G. (1965). A further study of the partial
reinforcement effect within-subjects. Psychonomic Science, 3, 533-534.

Perin, C. T. (1942). Behavior potentiality as a joint function of the amount of training and
the degree of hunger at the time of extinction. Journal of Experimental Psychology,
30, 93-113.






Quartermain, D., & Vaughan, G. M. (1961). Effect of interpolating continuous
reinforcement between partial training and extinction. Psychological Reports, 8,
235-237.

Rashotte, M. E. (1968). Resistance to extinction of the continuously rewarded response
in within-subject partial reinforcement experiments. Journal of Experimental
Psychology, 76, 206-214.

Rashotte, M. E., & Amsel, A. (1968). The generalized partial reinforcement effect:
Within S PRF and CRF training in different runways, at different times of day by
different experimenters. Psychonomic Science, 11, 315-316.

Rashotte, M. E., Ross, M., & Amsel, A. (1968). Generalization of the partial
reinforcement effect. Psychonomic Science, 11, 173-174.

Schmid, T. L. (1986). Reducing inappropriate behavior of mentally retarded children
through interpolated reinforcement. American Journal of Mental Deficiency, 91,
286-293.

Schmid, T. L. (1988). A comparison of two behavior reduction procedures: Traditional
extinction alone and interpolated reinforcement followed by extinction. Journal of
mental Deficiency Research, 32, 289-300.

Sidman, M., (1960). Tactics of scientific research: Evaluating experimental data in
psychology. New York: Basic Books, Inc.

Skinner, B. F. (1938). The behavior of organisms. Acton, MA: Copley Publishing
Group

Stalling, R. B., Moreland, J. W., Merrill, K. H., & Scotti, J. (1981). Continuous
reinforcement interpolated between intermittent reinforcement and extinction
decreases resistance to extinction. Behavior Analysis Letters, 1, 89-95.

Sutherland, N. S., Mackintosh, N. J., & Wolfe, J. B. (1965). Extinction as a function of
the order of partial and consistent reinforcement. Journal of Experimental
Psychology 69, 56-59.

Terrace, H. S. (1966). Stimulus control. In W. K Honig (Ed.) Operant Behavior: Areas
of Research and Application (pp. 271-344). New York: Appleton-Century-Crofts.

Theios, J. (1962). The partial reinforcement effect sustained through blocks of continuous
reinforcement. Journal of Experimental Psychology, 64, 1-6.

Walter, D. E., & Palya, W. L. (1986). Document set for experiment controllers with
ECBasic or ECL prom sets. Jacksonville, AL: William L. Palya.

Waters, W., & Knott, P. D. (1970). Tests of frustration theory extended to the
generalized partial reinforcement effect. Psychonomic Science, 20, 61-62.

Wickens, D. D., & Miles, R. C. (1954). Extinction changes during a series of
reinforcement-extinction sessions. Journal of Comparative and Physiological
Psychologv, 47, 315-317.




56

Williams, S. B. (1938). Resistance to extinction as a function of the number of
reinforcements. Journal of Experimental Psychology, 23, 506-522.

Wilson, M. P. (1954). Periodic reinforcement interval and number of periodic
reinforcements as parameters of response strength. Journal of Comparative and
Physiological Psychology, 47, 51-56.

Wylie, A. M., & Grossmann, J. A. (1988). Response reduction through the
superimposition of continuous reinforcement: A systematic replication. Journal of
Applied Behavior Analysis, 21, 201-206.












BIOGRAPHICAL SKETCH

Troy Zarcone was born in Biloxi, Mississippi, to Gail Scribner and Bart Zarcone

on December 19, 1965. He was an average student in the public education system. After

graduating high school in 1983, he enrolled at Indian River Community College where he

obtained his Associate of Arts degree.

In 1985, Troy was accepted to the University of Florida, Psychology Department.

He graduated in 1987 with his Bachelor of Science degree.

After graduation. Troy's future wife introduced him to the science of behavior by

giving him a copy of Walden Two by B.F. Skinner. After reading several other books by

B.F. Skinner, Troy returned to school in the summer of 1988 as a postbaccalaureate

student under the supervision of Dr. Brian Iwata. The fall of that year Troy took a class on

research methods taught by Dr. H. S. Pennypacker.

In the spring of 1989, Troy entered Graduate School in the Department of

Psychology and in the fall of 1991 received his master's degree.






I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and i ully adequate, in scope an quality,
as a dissertation for the degree of Doctor of Philo phy.


enry S. Pen ypacker, Cha an
Professor of Psychology

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.


Marc N. Branch
Professor of Psychology


I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Phil sophy.


ns van Haaren
sociate Scientist of Psychology


I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, scope and quality,
as a dissertation for the degree of Doctor of Philosophy.


rd IF. Malagdi
professor of Psychology

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.


Brian Iwata
Professor o sychology






I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosoph


(Ti nothy D.JIac'enberg
A instant Professor of Psychology


I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philg y.


Donald J. Seouwer
Associate Professor of Psychology


I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.


Cecil Merer
Professor of Special Education


This dissertation was submitted to the Graduate Faculty of the Department of
Psychology in the College of Liberal Arts and Sciences and to the Graduate School and
was accepted as partial fulfillment of the requirements of the degree of Doctor of
Philosophy.

December 1994
Dean, Graduate School























L
17:0


1z -


UNIVERSITY OF FLORIDA
I3 1262 08553 9186
3 1262 08553 9186