Relationship Between Contingency Awarene

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

(This is a sample cover image for this issue. The actual cover is not yet available at this time.

This article appeared in a journal published by Elsevier. The attached


copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy

Learning and Motivation 43 (2012) 55–65

Contents lists available at SciVerse ScienceDirect

Learning and Motivation


journal homepage: www.elsevier.com/locate/l&m

Relationship between contingency awareness and human


performance on random ratio and random interval schedules
Ceri A. Bradshaw ∗ , Phil Reed
Swansea University, UK

a r t i c l e i n f o a b s t r a c t

Article history: In three experiments, human participants pressed the space bar on a computer keyboard
Received 8 August 2011 to earn points on random-ratio (RR) and random-interval (RI) schedules of reinforcement.
Received in revised form Verbalized contingency awareness (CA) for each schedule was measured after the entire
18 November 2011
task (Experiments 1 and 2), or after each RR–RI trial (Experiment 3). In all three exper-
iments, an ability to accurately verbalize the appropriate contingency for each schedule
Keywords:
was associated with typical schedule responding. In addition, lagged correlational anal-
Variable ratio
Variable interval
ysis of verbalized contingency awareness and schedule performance was undertaken in
Schedule of reinforcement Experiment 3. Results of these correlations initially suggested that CA scores were bet-
Language ter predictors of typical schedule performance on the following trial than typical schedule
Contingency awareness performance was of CA scores on the following trial; however, further analysis revealed
Humans no significant differences between the alternate lagged correlation types in each pair. Find-
ings, therefore, suggest that operant performance on RR and RI schedules is associated with
participant-derived rule governed behavior, and results are discussed in terms of the intro-
duction of a new paradigm for measuring this association, and the potential for further
research to manipulate these parameters further in order to investigate causality in this
respect.
© 2012 Elsevier Inc. All rights reserved.

Variable ratio (VR) and random ratio (RR) schedules support higher rates of response than variable interval (VI) or random
interval (RI) schedules that provide comparable rates of reinforcement; this is a finding that holds across many different
species (see Catania, Matthews, Silverman, & Yohalem, 1977; Catania & Reynolds, 1968; Ferster & Skinner, 1957; Peele,
Casey, & Silverberg, 1984; Reynolds, 1975; Zuriff, 1970). This VR/RR versus VI/RI schedule response rate difference has been
the subject of much theoretical debate with respect to learning (see Reed, 2006). Although such response rate differences
between VR/RR and VI/RI schedules are also seen in human participants (see Baron & Galizio, 1983; Matthews, Shimoff,
Catania, & Sagvolden, 1977; Raia, Shillingford, Miller, & Baier, 2000; Reed, 1993, 1994, 1999, 2001), such nonhuman typical
behavior is manifested in humans often as a consequence of the employment of complex tasks, and/or attempts to prevent
interference from verbalization in the completion of tasks (see Lowe, Harzem, & Bagshaw, 1978; Lowe, Harzem, & Hughes,
1978; Matthews et al., 1977). However, the fact that human responding on schedules of reinforcement can show patterns
that differ from nonhuman performance is thought to be of some theoretical importance in understanding the factors that
control schedule behavior in humans (e.g. Leander, Lippman, & Meyer, 1968; Lippman & Meyer, 1967; Lowe, 1979; Lowe,
Harzem, & Hughes, 1978; Weiner, 1964, 1969, 1970).

∗ Corresponding author at: Department of Psychology, Swansea University, Singleton Park, Swansea SA2 8PP, UK.
E-mail addresses: 149987@swansea.ac.uk (C.A. Bradshaw), p.reed@swansea.ac.uk (P. Reed).

0023-9690/$ – see front matter © 2012 Elsevier Inc. All rights reserved.
doi:10.1016/j.lmot.2011.11.002
Author's personal copy

56 C.A. Bradshaw, P. Reed / Learning and Motivation 43 (2012) 55–65

Of course, differential responding to schedules by humans compared to nonhumans may be due to a wide variety of factors,
many of which are procedural in nature. For example, the type of reinforcement employed (see Lowe, Harzem, & Bagshaw,
1978), the delivery style of reinforcers (Matthews et al., 1977), and the physical exertion demanded by the experimental task
(Matthews et al., 1977) may all impact on schedule performance. However, of particular theoretical interest is the suggestion
that these differences may be due to the capacity for verbal behavior in humans (see Baron & Galizio, 1983; Baron & Perone,
1982; Harzem, Lowe, & Bagshaw, 1978; Lowe, Beasty, & Bentall, 1983; Skinner, 1969). Verbal behavior is suggested as a key
factor promoting species differences, in that differences between human and nonhuman schedule performance can follow
direct verbal instructions given to humans (e.g. Baron, Kaufman, & Stauber, 1969; Harzem et al., 1978; Kaufman, Baron, &
Kopp, 1966; Lippman & Meyer, 1967; Matthews et al., 1977; Shimoff, Catania, & Matthews, 1981; Weiner, 1970), or are
associated with changes to the verbalizations made by humans while concurrently completing a schedule of reinforcement
task (Laties & Weiss, 1963; Lowe, Harzem, & Bagshaw, 1978).
Further evidence for this suggestion comes from experiments with human infants and children that have demonstrated
that operant behavior (typically on fixed interval, FI, schedules) in infants who are yet unable to talk, is almost identical to that
of nonhuman subjects (Lowe et al., 1983). In contrast, young children who have acquired some language skills, but who have
not developed full functional language, respond with patterns that are differentiated from non-humans, human infants, and
human adults (Bentall, Lowe, & Beasty, 1985); older children, with improved language skills, exhibit response patterns typical
of human adults (Bentall et al., 1985; Bentall & Lowe, 1987). It is suggested that the verbal rules about the schedules that
are generated by the participants, especially when elicited by the direct instructional elements of experimental procedure,
are rigidly adhered to at the expense of responding sensitively to the contingencies of the schedule (see Bentall et al., 1985;
Bentall & Lowe, 1987; Lowe, 1979; Lowe et al., 1983). This research is informative in presenting a possible explanation for
the differences between the operant behavior observed in humans and non-humans.
Evidence from classical conditioning also suggests that verbalizations may play an important role in human performance
in conditioning experiments, as many studies indicate that only those human participants who show awareness (often
expressed verbally) of the contingency of the UCS–CS pairings show conditioned responding (see Brewer, 1974; Dawson &
Schell, 1985; Mitchell, De Houwer, & Lovibond, 2009, for reviews of classical conditioning and contingency awareness). It
should be noted that much of the research that also supports this view derived from instrumental conditioning (e.g. Bentall
et al., 1985; Bentall & Lowe, 1987; Lowe et al., 1983) is based on experimental procedures involving FI schedules, which do
not generate patterns of responding similar to those seen on VR/RR and VI/RI schedules (see Ferster & Skinner, 1957). Given
the theoretical importance of the VR/RR versus VI/RI schedule difference, and the potential theoretical importance of verbal
behavior for human performance, it would seem important to further investigate the influence of the latter in the context
of RR and RI schedules.
Some research has been conducted on the relationship between VR/RR and VI/RI schedules and verbal behavior. For
example, the effects of contingency-governed versus instructed learning on responding on VR and VI schedules has received
some attention (see Catania, Matthews, & Shimoff, 1982; Matthews et al., 1977; Shimoff et al., 1981). These studies, typically,
have demonstrated that contingency shaping can produce VR and VI responding in humans akin to that of non-humans,
especially when instructions are minimal, and a consummatory response to the reinforcer is required (see Matthews et al.,
1977; Raia et al., 2000). Other studies have clarified that verbally instructed responding on VR/RR schedules is relatively
insensitive to changes in the actual contingency, even when responding patterns achieve low rates of reinforcement (e.g.
Catania, Lowe, & Horne, 1990; Catania et al., 1982; Matthews, Catania, & Shimoff, 1985; Shimoff et al., 1981; Shimoff,
Matthews, & Catania, 1986).
However, the clearly complex relationship between verbal behavior and human operant performance on VR/RR and VI/RI
schedules is yet to be determined with any clarity (see Harzem et al., 1978; Lowe, 1979; see Hayes, 1989 for a review). Thus,
the simple aim of the present series of studies was to consider patterns of responding in human participants exposed to RR
and RI schedules in light of previous findings in relation to impact of verbal behavior on operant performance. In particular,
the role of verbalized contingency-awareness (see Catania et al., 1982; Matthews et al., 1985; Shimoff et al., 1986) was
examined in terms of its relationship with schedule performance.

Experiment 1

As noted above, VR or RR schedules generally produce higher rates of responding than VI or RI schedules in nonhumans
(Ferster & Skinner, 1957; Peele et al., 1984), and also in human participants (Catania et al., 1982; Matthews et al., 1977;
Shimoff et al., 1981). However, it has been shown that responding according to schedule type in humans is influenced by
the presence of verbal rules that correspond to the contingency being experienced (see Lowe, 1979; Lowe et al., 1983;
Matthews et al., 1977; Raia et al., 2000). In general, if a verbal rule deriving from instructions provided by the experimenter
is accurately reported, then performance on VR/RR and VI/RI schedules resembles that of non-humans, but the effects of
inaccurate reports lead to very different patterns of responding, which are also remarkably resistant to change (see Hayes
et al., 2010, for a review).
The present study aimed to examine the relationship between responding on RR schedules and participants’ ability to
accurately describe the contingencies (contingency awareness) on ratio and interval schedules. It was hypothesized that
subjects who show greater contingency awareness for the appropriate rule of reinforcement according to schedule would
Author's personal copy

C.A. Bradshaw, P. Reed / Learning and Motivation 43 (2012) 55–65 57

demonstrate more schedule-typical patterns of responding (i.e. faster rates of responding for RR schedules and slower rates
of responding for RI schedules). This hypothesis has not been tested previously.

Method

Participants
A sample of 18 students was recruited via the Psychology Department subject-pool system, and they received credit for
their participation (but no financial payment). The sample comprised 15 females and 3 males, aged between 19 and 28 years
(mean = 21.0 ± 2.1).

Apparatus
The experimental task was presented on a standard desktop computer. Visual Basic (6.0) was used to program the task,
which consisted of two primary reinforcement schedule types.
The program firstly presented a random ratio (RR) schedule, wherein points (40 points), acting as reinforcers, were
awarded for presses to the space bar according to a RR-10 schedule (points were awarded after each space bar response with
a 1/10 probability). Participants also lost one point for each space bar response, regardless of whether the hit was reinforced.
This manipulation was introduced as it has previously been noted that, in the absence of cost for a key press, there is no
penalty for over-responding, and, therefore, no reason to regulate performance in line with the contingency of the schedule
(see Reed, 2001; Shanks & Dickinson, 1991).
Secondly, an RI schedule was presented, whereby 40 points were awarded following the first response after a specified
amount of time had elapsed. The RI schedule was yoked to the preceding RR schedule, so that reinforcement in the RI
schedule was delivered only after the elapse of the time that it had taken for the corresponding reinforcer to be awarded on
the RR trial.
The computer task presented a white screen, with three stimulus boxes; one placed in the center upper portion of the
screen, one in the center middle portion of the screen, and one in the center lower portion of the screen. The top box was
approximately 8 cm wide × 3 cm high, and was blocked with a single color (either red or green), to indicate schedule type. A
new schedule was, thus, indicated by the changing color of this box. For the first trial (RR) it was red (for half the participants),
followed by green for the second trial (RI), and alternated in this manner for the subsequent trials. For the other half of the
participants the box was green for the first (RR) schedule and red for the second (RI) schedule, with this pattern recurring
over subsequent trials. Participants were informed that the box would change color when a new trial commenced, but not
informed of which schedule type the color indicated. The lower box was also approximately 8 cm wide × 3 cm high, and
stated “click here”. The participants were requested to place the mouse cursor in this box, and then use the space bar to earn
points. The middle box contained the running points total, with the word “points” situated next to it. Points were set to 10 at
the commencement of the task, and reduced by one with every space bar press. When reinforcement was delivered, points
would rise by 40.

Materials
A contingency awareness questionnaire was developed to retrospectively measure the participants’ awareness of rules
governing reinforcement (through points awarded) in the task. This questionnaire was based on the questions used by
Catania et al. (1982; see also Matthews et al., 1985; Shimoff et al., 1986) to assess contingency awareness, with the notable
difference that participants in the afore-mentioned research were required to fill in the answer to the question in their own
words. In the present research, the question posed in each of the questionnaire items was based on the questions asked in
the research cited above, but the participants were required to indicate their answer in terms of one closed responses from
seven options presented for each question. In total eight questions were asked, each related to one of the eight trials within
the task, and each item asked the same question: “In the [first/second] game in the task, what did you consider to be the best
way to increase your points score?”: (a) Press space bar frequently; (b) Press space bar very little; (c) Do not press space bar at all;
(d) Press space bar a lot in a small amount of time; (e) Pause in-between space bar presses; (f) Press space bar a little in a small
amount of time; (g) I do not know.
For RR trials, the correct response was deemed to be option (d) “Press space bar a lot in a small amount of time”. If
participants indicated this response for the items related to RR trials (1, 3, 5, and 7) they received a score of 2 points for each
correct response (after Catania et al., 1982; Matthews et al., 1985; Shimoff et al., 1986). However, option (a) “Press space
bar frequently” was considered to be partially correct, and, for this reason, responses indicating this option in relation to RR
trials were given a score of 1 (after Catania et al., 1982; Matthews et al., 1985; Shimoff et al., 1986). For RI trials, the correct
response was considered to be option (e) “Pause in-between space bar presses”. Participants were awarded 2 points each time
they indicated this response in relation to RI trials (2, 4, 6, and 8; after Catania et al., 1982; Matthews et al., 1985; Shimoff
et al., 1986). Any other responses indicated received a score of zero. Thus, the maximum score it was possible to be awarded
for contingency awareness across all eight trials was 16.

Procedure
Participants were tested individually in a quiet room, which contained a desk and computer, with the monitor situated
approximately 60 cm from them. Participants gave written consent, and read the study information and paper instructions
Author's personal copy

58 C.A. Bradshaw, P. Reed / Learning and Motivation 43 (2012) 55–65

250

RR Low CA
200

Mean Response Rates (ms)


RR High CA
RI Low CA
RI High CA
150

100

50

0
1 2 3 4
Trial

Fig. 1. Experiment 1 mean response rates per minute – RR and RI trials for higher and lower contingency awareness groups.

for the task. Participants commenced the task in their own time and were required to fill in basic demographic details about
themselves before the schedule task was presented.
Each schedule presentation was 4 min long, and a RR schedule trial was always presented immediately prior to the yoked
RI schedule trial. There were four presentations of a yoked RR–RI pair. The procedure of yoking RI trials to preceding RR trials
ensured that reinforcement in the RI schedule was delivered after a time interval equal to that taken for the corresponding
reinforcer to be awarded on the RR trial.
Prior to the task beginning, participants were presented with the following instructions on the computer screen:
“When the task begins, position the cursor over the rectangle which states “click here”, and use the space bar to score as
many points as possible. There are eight games in total. The first game is identified with a large red [green] rectangle at
the top of the screen. When the first game is over, the rectangle will change to green [red] to indicate the start of the next
game. The rectangles alternate between red and green [green and red] to indicate the changing games for the remainder of
the task. Your goal in each game is to reach the highest score possible. You will see that the points reduce according to the
way in which you play, but will rise again every so often, according to the pattern of space bar hits that you use. All you
need to do is to find the best pattern of space bar hits to score as highly as possible in each game. It may be a good idea to
respond quickly sometimes, and slowly at other times, but you need to discover this for yourself!”
After task completion, participants completed the contingency questionnaire before leaving the laboratory. Participants
were then debriefed and paid in subject-pool credit.

Results and discussion

Summed contingency awareness (CA) scores for every trial ranged from 0 to 16 across the sample, with a mean of 8.1
(±4.6). The sample was split into two groups according to contingency awareness score. This was done by calculating the
mean CA score across all three experiments reported in this series, and then splitting the participants into two groups, above
and below the mean, using the same values for each experiment. Using a CA mean across all three experiments ensured that
Higher and Lower CA groups would be comparable with each other between experiments. The mean of all CA scores over
all three experiments was 7.7; consequently, scores in the lower half of the possible range (0–7) were considered to have
relatively lower contingency awareness in all three experiments, while scores in the higher half of the possible range (8–16)
were described as showing higher contingency awareness. For this experiment, Group Lower consisted of 7 participants
(mean CA score = 3.1 ± 2.0), while Group Higher consisted of 11 participants (mean CA score = 11.2 ± 2.5).
Fig. 1 presents the response rates for each exposure to each of the two schedule types, for the two groups of participants
(Lower and Higher). Inspection of these data reveals that participants in the higher contingency awareness group (Higher
CA) displayed higher rates of response on the RR than on the RI schedule. In contrast, the lower contingency awareness
(Lower CA) group displayed little difference between responding on the RR and RI schedules across the trials.
A three-factor, mixed-model analysis of variance (ANOVA) with group as a between subject factor, and schedule and
trial as within-subject factors was conducted on these data and showed a statistically significant main effect of schedule,
F(1,16) = 39.65, p < 0.001, but no statistically significant main effects of trial, p > 0.30, or group, F < 1. There was a statistically
significant interaction between group and schedule, F(1,16) = 16.04, p < 0.001, but no other statistically significant interac-
tions, all ps > 0.30. Simple effect analyses of schedule for Higher and Lower CA groups conducted on the response rate data
Author's personal copy

C.A. Bradshaw, P. Reed / Learning and Motivation 43 (2012) 55–65 59

averaged across the four trials revealed a statistically significant effect for RR versus RI schedule for the Higher CA group,
F(1,22) = 278.836, p < 0.001, but no statistically significant effect for the Lower CA group, p > 0.05
These findings suggest that humans can display the typical response rate difference between the RR and RI schedule
types as observed in non humans (see Catania & Reynolds, 1968; Catania et al., 1977; Ferster & Skinner, 1957; Peele et al.,
1984; Reynolds, 1975; Zuriff, 1970). These data replicate previous findings on humans (e.g. Baron & Galizio, 1983; Lowe,
Harzem, & Hughes, 1978; Lowe, Harzem, & Bagshaw, 1978; Matthews et al., 1977; Reed, 1993, 1994, 1999, 2001). However,
importantly, the above RR–RI response rate difference was only strongly evident in participants with higher contingency
awareness. This finding is in line with previous studies that have focused on the relationship between verbal behavior and
schedule performance (see Bentall et al., 1985; Bentall & Lowe, 1987; Catania et al., 1982; Hayes et al., 1989; Lowe et al.,
1983; Matthews et al., 1977; Shimoff et al., 1981; Vaughan, 1989) and suggests that the formation of verbal rules is associated
with schedule behavior in human responding on RR and RI schedules.

Experiment 2

In order to extend the generality of the results from Experiment 1, it was deemed prudent to replicate the study using a
different ratio value. To this end, a RR-30 schedule was employed in Experiment 2.

Method

Participants and apparatus


A sample of 18 students was recruited via the Psychology Department subject-pool system as described for Experiment
1. The sample comprised 14 females and 4 males, aged between 19 and 29 years (mean = 20.5 ± 2.7). The apparatus was
identical to that described in Experiment 1.

Procedure
The procedure was as described in Experiment 1. Participants were tested individually, in a quiet room. Each schedule
presentation was four minutes long, and a RR schedule trial was always presented immediately prior to the yoked RI schedule
trial. Each RR–RI pair was presented four times. The RR value was 30 (each response had a 1/30 probability of reinforcement),
and yoking to the RI was conducted as described in Experiment 1.

Results and discussion


Summed contingency awareness scores across the trials ranged from 0 to 16, with a mean of 5.4 (±4.9, SD). The sample
was split into two groups according to their contingency awareness score as described in Experiment 1; a score of 0–7 was
classed as Lower CA, while a score of 8–16 was classed as Higher CA. Group Lower consisted of 12 participants (mean CA
score = 2.5 ± 2.1), and Group Higher consisted of 6 participants (mean CA score = 11.3 ± 3.2).
Mean response rates for each of the two schedule types, RI and RR for both Lower and Higher CA groups, can be seen
in Fig. 2. In accordance with findings from Experiment 1, participants in the Higher CA group displayed higher rates of
response on the RR compared to the RI schedule. The Lower CA group appeared to show a much smaller difference between
responding on the RR and RI schedules across the trials than the Higher CA group.
These data were analyzed by a three-factor, mixed-model ANOVA (group × schedule × trial), which revealed a statistically
significant main effect of schedule, F(1,15) = 7.15, p < 0.05, but no statistically significant main effects of group or trial, Fs < 1.
There was a statistically significant interaction between group and schedule, F(1,15) = 4.58, p < 0.05, but no other statistically
significant interactions, all ps > 0.60. Simple effect analyses of schedule for Higher and Lower CA groups conducted on the
averaged data across all four trials revealed a statistically significant effect for RR versus RI schedule for the Higher CA group,
F(1,22) = 16.05, p < 0.01, but no statistically significant effect for the Lower CA group, F < 1.

200
Mean Response Rates (ms)

180
160 RR Low CA
140 RR High CA

120 RI Low CA
RI High CA
100
80
60
40
20
0
1 2 3 4
Trial

Fig. 2. Experiment 2 mean response rates per minute – RR and RI trials for higher and lower contingency awareness groups.
Author's personal copy

60 C.A. Bradshaw, P. Reed / Learning and Motivation 43 (2012) 55–65

Thus, Experiment 2 yielded similar findings to those of Experiment 1, in that a difference can be observed for RR versus
RI schedules for participants with higher contingency awareness, but not for those with lower contingency awareness.

Experiment 3

The findings from Experiments 1 and 2 demonstrated a higher response rate in RR schedules than in RI schedules in higher
contingency aware participants for two different schedule parameters, RR-10 and RR-30. However, both of these experiments
used an entirely retrospective measure of contingency awareness, in that the contingency awareness questionnaire for all
trials was administered at the end of the whole task. It is possible that shaping across trials influenced this retrospective
contingency awareness; appropriate shaped responding in later trials led to greater levels of reinforcement, and allowed
Higher CA participants to hazard more accurate guesses regarding the correct reinforcement rules for all trials. The final
experiment aimed to investigate contingency awareness directly after exposure to each trial, and also to attempt to assess
the extent to which schedule performance precedes the verbalization (spoken or unspoken) of the appropriate rule for the
schedule, or vice versa.
As noted in the General Introduction, previous studies have established that verbal rules play an important role in human
schedule performance (see Hayes, 1989 for a review), and a relationship between accurate verbalization of the correct
contingency for a schedule and performance on that schedule also has been demonstrated. However, the precise nature of this
relationship is yet to be established. Catania et al. (1982), and Lowe et al. (1983) have suggested that schedule performance
follows accurate verbalization of a rule (see Bentall et al., 1985; Bentall & Lowe, 1987; Catania et al., 1982; Lowe et al.,
1983; Shimoff et al., 1986). Despite the research interest in this area, however, it has not been successfully demonstrated
that verbalization of a correct contingency precedes appropriate performance on the subsequent schedule. It may also
be that appropriate performance on a schedule could lead to accurate subsequent verbal reporting of the contingency
in retrospect. In other words, individuals are better able to verbalize a contingency after they have been reinforced for
contingency-appropriate behavior.
Given this, the aim of the final experiment was to replicate the previous two experiments, with one important difference
in terms of the measurement of contingency awareness. Rather than measure this retrospectively for every schedule at
the end of the whole task, contingency awareness was measured at the end of each trial. It was also hoped to correlate
the CA measure with both the previous, and the subsequent, schedule performance. Higher correlations between CA and
appropriate performance on the subsequent trial would suggest that rule formation precedes appropriate performance;
whereas higher correlations between CA and appropriate performance on the previous trial would suggest that appropriate
performance governs accurate rule formation.

Method

Participants and apparatus


A sample of 50 students was recruited via the Psychology Department subject-pool system as described in Experiment
1. The sample comprised 25 females and 25 males, aged between 19 and 25 years (mean = 21.8 ± 1.9). The apparatus and
materials were as described in Experiment 1, with the exception that no contingency awareness questionnaire was employed.

Procedure
Participants were tested individually as described in Experiment 1. Each schedule presentation was four minutes long,
and a RR-30 schedule trial was always presented immediately prior to the yoked RI schedule trial. Each RR–RI pair was
presented four times, and the schedule values were determined as described in Experiment 1.
The primary variation to this experiment in comparison to Experiments 1 and 2 was that, rather than being asked to
complete a contingency awareness questionnaire retrospectively for each trial at the end of the entire task, participants
paused after each trial and stated aloud, in their own words, what they thought best described the contingency governing
the task. The question asked was derived from previous research into participants’ ability to understand the schedule to
which they had just been exposed (e.g. Catania et al., 1982; Hayes, Brownstein, Zettle, Rosenfarb, & Korn, 1986; Matthews
et al., 1985; Shimoff et al., 1986). This question asked: “What was the best approach to gaining points in the previous trial?”
The participants then stated what they believed to be the best way of scoring points for that trial, and this response was
recorded by the experimenter.
These responses were later subjected to a manifest content analysis. The phases of the content analysis employed were
conducted in line with well-established procedures used and recommended by Osborne and Reed (2008) and Vaughn,
Schumm, and Sinagub (1996). If the participant’s response accurately described the preceding contingency, then two points
were scored. If the participant’s response was thought to partially describe the preceding contingency, then 1 point was
scored. Two independent raters completed this procedure to verify the reliability of the coding. A Cohen’s kappa analysis
was used to check the inter-rater reliability. A high mean level of reliability (0.81) was identified between their separate
judgments of the participants’ responses.
Author's personal copy

C.A. Bradshaw, P. Reed / Learning and Motivation 43 (2012) 55–65 61

250

Mean Response Rates (ms)


200

RR High CA
RR Low CA
150
RI High CA
RI Low CA

100

50

0
1 2 3 4
Trial

Fig. 3. Experiment 3 mean response rates per minute – RR and RI trials for higher and lower contingency awareness groups.

Results and discussion


Summed contingency awareness (CA) scores across the sample ranged from 0 to 16, with a mean of 9.6 (±5.3). The sample
was split into two groups according to their CA score, as described before; a score of 0–7 was classed as Lower CA, while a
score of 8–16 was classed as Higher CA. Group Lower CA consisted of 12 participants (mean CA score = 1.7 ± 2.6), and Group
Higher CA consisted of 38 participants (mean CA score = 12.4 ± 2.9).
Fig. 3 displays mean response rate across schedules and trials for both groups. The discrepancy in rates of response
between schedule types for participants in the Higher CA group (RR rates being higher than RI rates) appears to be greater
than for those in the Lower CA group, whose response rates did not differ very much between the schedule types.
These data were analyzed by a three-factor, mixed-model ANOVA (group × schedule × trial), which revealed statisti-
cally significant main effects for schedule, F(1,48) = 13.13, p < 0.001, trial, F(3,144) = 4.27, p < 0.01, and group, F(1,48) = 7.86,
p < 0.01. There were also statistically significant interactions between group and schedule, F(1,48) = 9.50, p < 0.01, schedule
and trial, F(3,144) = 4.65, p < 0.01, and a statistically significant three-way interaction, F(3,144) = 2.67, p < 0.05, but there was
no statistically significant interaction between trial and CA group, p > 0.135.
Follow-up treatment of data, using two-factor ANOVAs (schedule × trial) for each group, indicated that, for Lower CA
participants, there was no statistically significant main effects of schedule or trial, Fs < 1, and no statistically significant
interaction between schedule and trial, F < 1. For Higher CA participants, there were statistically significant main effects
of schedule, F(1,37) = 36.45, p < 0.001, and trial, F(3,111) = 9.78, p < 0.001, and a statistically significant interaction between
schedule and trial, F(3,111) = 11.95, p < 0.001.
These results demonstrate a clear relationship between schedule type and response rate for those participants who can
accurately describe the contingency, but not for those who could not describe the contingency as accurately. These findings
further support the previous findings that humans can demonstrate similar patterns of responding to non-humans (e.g.
Baron & Galizio, 1983; Lowe, Harzem, & Bagshaw, 1978; Lowe, Harzem, & Hughes, 1978; Matthews et al., 1977), but the
present findings also confirm those of the previous two experiments, which suggest that this difference is only apparent in
participants with relatively higher contingency awareness.
Lagged Pearson correlations were performed between the CA scores for each successive RR plus RI schedule pair (CA for
the RR trial plus CA score for the corresponding yoked RI trial), and the response rate difference score for each pair of yoked
schedules (RR minus RI). The former score gave an estimate of contingency awareness for a pair of trials, and the latter gave
an indication of schedule appropriate behavior (i.e. RR responding being greater than RI responding). Lagged correlations
were performed between CA scores for an earlier trial and response rate difference scores for the subsequent trial. These
revealed a statistically insignificant correlation between CA scores for Trial Pair 1 and response rate difference scores for
Trial Pair 2 (r = .248, non-sig). A statistically significant correlation was found between CA scores for Trial Pair 2 and response
rate difference scores for Trial Pair 3 (r = .439, p < 0.001), and a further statistically significant correlation was found between
CA scores for Trial Pair 3 and response rate difference scores for Trial Pair 4 (r = .639, p < 0.001). Lagged Pearson correlations
were also performed between response rate difference scores for an earlier trial and CA scores for the subsequent trial. A
statistically insignificant correlation was found between response rate difference scores for Trial Pair 2 and CA scores for Trial
Pair 3 (r = .220, non-sig). A statistically significant correlation was found between response rate difference scores for Trial
Pair 3 and CA scores for Trial Pair 4 (r = .385, p < 0.01), and a further statistically significant correlation was found between
the response rate difference scores for Trial Pair 3 and CA scores for Trial Pair 4 (r = .463, p < 0.001).
The results of the correlations between contingency awareness for each trial and difference scores between the responses
on schedule types for the subsequent or preceding trial numerically demonstrated stronger relationships between contin-
gency awareness scores for one trial and difference scores for the next trial than for contingency awareness scores for one
Author's personal copy

62 C.A. Bradshaw, P. Reed / Learning and Motivation 43 (2012) 55–65

trial and difference scores for the previous trial. This observation could suggest that there is a stronger relationship between
contingency awareness on the preceding trial and schedule-appropriate responding on the next trial than for schedule-
appropriate responding on one trial and contingency awareness on the next trial. However, when analyzed statistically,
there were no significant differences revealed between the r values for alternate lagged correlation types in each pair. The
difference between the correlation between CA Trial Pair 1 and Response Rate Difference Pair 2 and the correlation between
Response Rate Difference Pair 1 and CA Trial Pair 2 was z = .14, p = 0.89 for a two-tailed hypothesis. The difference between
the correlation between CA Trial Pair 2 and Response Rate Difference Pair 3 and the correlation between Response Rate
Difference Pair 2 and CA Trial Pair 3 was z = .32, p = 0.75 for a two-tailed hypothesis. The difference between the correlation
between CA Trial Pair 2 and Response Rate Difference Pair 3 and the correlation between Response Rate Difference Pair 2
and CA Trial Pair 3 was z = 1.24, p = 0.22 for a two-tailed hypothesis.
Thus, it is safest to conclude that these findings do not demonstrate conclusively that participants with higher contin-
gency awareness were formulating a verbal rule for the contingency of the schedule and then acting appropriately to their
rule, rather than allowing their schedule performance to determine their verbalization of a rule to define it. Further explo-
ration, perhaps parametric in nature, would be needed to determine these relationships. However, the current study did
demonstrate the utility of a procedure that could shed some light on this topic.

General discussion

The present series of studies was designed to investigate the performances of human participants on RR and RI schedules,
with special reference to the role of contingency awareness, as measured by self-report. The findings of all three experiments
indicate that humans can show typical non-human schedule response rates, with marked differences in the performance
between RR and RI schedules; response rates were higher on the RR compared to the RI schedule. These findings are in line
with those presented by Raia et al. (2000), Randell, Ranjith-Kumar, Gupta, & Reed (2009), and Reed (2001). However, the
current studies all demonstrated that not all humans show these schedule responding differences. In fact, only individuals
who were able to verbalize the correct contingency for the schedule parameter showed this appropriate schedule responding.
In addition to demonstrating a relationship between schedule performance and verbalized contingency awareness, Exper-
iment 3 attempted to measure whether verbalized contingency awareness preceded schedule-typical performance, or vice
versa. However, the lagged-correlation findings from this study were inconclusive and did not adequately demonstrate that
contingency awareness and schedule performance operate in a clear, linear direction. Thus, it cannot be concluded that
participants formulated a verbal rule to describe the schedule contingency first and then performed according to that rule,
or vice versa. Development of further measures of contingency awareness to utilize in conjunction with schedule tasks, and
new experimental paradigms for their investigation, may offer opportunities for enhanced clarification of this pivotal matter
of theoretical concern.
It should be noted that the mean contingency awareness scores varied across the three experiments presented; for
example, the mean CA score for Experiment 3 was considerably higher than for Experiment 2. The difference between these
scores was not analyzed statistically due to the many differences in procedure between these studies; it is possible that such
procedural differences between Experiments 2 and 3 (verbalizing CA at the end of the entire task, rather than at the end of
each RR–RI trial, respectively) could explain the difference in mean CA scores between the experiments. Again, this finding
could be further explored in future investigations.
Although there are studies that have suggested that classical conditioning can occur without awareness, most evidence
from classical conditioning paradigms shows that contingency awareness is necessary for putative conditioning effects to be
observed. Many studies have demonstrated that only participants who are aware of the pairing contingency show a response
to a conditioned stimulus (see Brewer, 1974; Dawson & Schell, 1985; Mitchell, De Houwer, & Lovibond, 2009 for reviews of
the literature). On this basis, it might be concluded that the available evidence from a wide range of conditioning procedures
supports the view that awareness accompanies conditioning, and that findings that do not show this link are likely to be
confounded by insensitive measures of contingency awareness.
Recent work on conditioning and awareness has focused upon evaluative conditioning (EC), which refers to a process
whereby a neutral conditioned stimulus (CS), which is paired with a highly attractive or aversive stimulus (the unconditioned
stimulus; UCS), adopts the positive or negative characteristics of the UCS and elicits the same reaction. The process may not be
entirely identical to classical conditioning, as it is thought to involve a cognitive component or evaluation of the UCS; hence,
evaluative conditioning. The attitude formed through the EC process is also generally very stable and resistant to extinction,
unlike classically conditioned responses. Such EC research is relevant to human schedule performance (see Dack, McHugh,
& Reed, 2010), but the EC literature is rather mixed in regard to the impact of awareness; some studies show conditioning
without awareness (e.g. Baeyens, Eelen, & Van den Bergh, 1990; Dickinson & Brown, 2007; Walther & Nagengast, 2006), and
some show the traditional link between conditioning and awareness (e.g. Pleyers, Corneille, Luminet, & Yzerbyt, 2007; Stahl
& Unkenbach, 2009; Stahl, Unkelbach, & Corneille, 2009). Although the relationship between conditioning and awareness
does not always hold true for EC, a meta-analysis of findings in the area indicated that evaluative conditioning effects were
stronger for higher rather than lower contingency aware participants, and that contingency awareness was a key moderator
of EC (Hofmann, De Houwer, Perugini, Baeyens, & Crombez, 2010).
The acknowledgement that a verbal rule constructed by the individual may act as a controlling stimulus in human
operant behavior was suggested by Skinner (1966), and, although the relationship between verbal behavior and schedule
Author's personal copy

C.A. Bradshaw, P. Reed / Learning and Motivation 43 (2012) 55–65 63

performance in humans has been investigated in a variety of ways, the precise direction of this relationship remains
indistinctly verified. The research previously undertaken by researchers in this field has often focused on schedule
responding driven by experimenter-given rules, or by shaping alone, but, as Hayes et al. (1986) have pointed out, the
differences in performance in these two scenarios could implicate differences in the social contingencies surrounding
verbalized rules that are either self-generated, or that have been declared by others. Although differences in performance
could be due to a simple effect of participants responding sensitively to the contingency of the schedule versus responding
blindly to the instructions offered by experimenters, they could also be due to differentially socially reinforced reasons for
following instructions, either those self-generated, or those presented by the researcher (Hayes et al., 1986). Therefore, any
associations between human schedule performance and self-governed rule formation, versus experimenter-governed rule
formation, must be investigated under carefully controlled conditions to establish subtle differences in response-patterns
which may have been overlooked in previous research in this area.
Difficulties with accurately measuring verbal behavior, in this case in the form of contingency awareness, are well doc-
umented (e.g. Ericsson & Simon, 1980; Hayes, 1986; Hayes & Hayes, 1992; see Hayes, 1989 for a review). The current
study operationalized two measures of verbal contingency awareness, one in the form of closed responses on a pre-printed
questionnaire and the other in the form of freely verbalized responses by participants which were recorded and coded
by experimenters. Generality of measures of verbalized contingency awareness is therefore increased somewhat by the
use of two different forms; however, we note that the validity of measurement of contingency awareness requires further
examination.
The question of why contingency-sensitive performance, as seen in non-human animals, is shown only in humans with
high contingency awareness is not answered by the research in the present study. Nevertheless it remains a critical theoret-
ical issue. One explanation for this phenomenon is that verbalizations in humans may interfere with contingency-shaped
schedule responding as demonstrated in non-humans. If this were the case, the interference experienced could mani-
fest in a number of ways; for example, it could disturb the trial-and-error process that ultimately leads to development
of contingency awareness in some, but not all participants. Or it could be that all humans will form hypotheses about
how their responding influences reinforcement; in some humans, the hypothesis developed leads to contingency-sensitive
responding, but in others it leads elsewhere. One way of measuring whether humans with lower contingency awareness
have developed inaccurate hypotheses regarding reinforcement is to consider their responses on measures of contin-
gency awareness. If they repeatedly identify the same incorrect response as being the most efficient way to earn points,
an incorrect hypothesis may have been set. Their responses to the CA measure could be cross-validated with data from
their patterns of precise responding on each trial to examine any consistency between their false hypothesis and schedule
performance.
The current study attempted to measure self-generated rules that were articulated by the participants and did not attempt
to manipulate schedule performance by direct instructional control formulated by the experimenter; this is doubtless an
important focus for future work which should be undertaken in this area.
Schedule performance in relation to these two scenarios requires investigation as any differences between the two
may hold important implications for human schedule performance. In particular the current study has introduced a new
methodology for research into the area of verbalized contingency awareness and schedule performance in humans, which
may be amended or extended to accommodate new measures or procedures for measuring the two factors in relation
to each other. It is hoped that this new methodology may lead to increased validity in measures of both verbalized
contingency awareness and schedule performance in humans and that causal relationships might be established with
time.
In summary, the primary findings of the present series of studies are that distinct ratio versus interval schedule response
rates can be observed in human participants, but that this effect is only observed in higher contingency aware individuals.
These results are presented as useful baseline findings generated from a methodology which could provide a paradigm
for further investigation into the complex relationships between verbalization, rule-governed behavior and schedule
responding.

Acknowledgment

Many thanks to Jordan Randell and Gary Freegard for help with computer programming for this study.

References

Baron, A., & Galizio, M. (1983). Instructional control of human operant behaviour. Psychological Record, 33, 495–520.
Baron, A., Kaufman, A., & Stauber, K. A. (1969). Effects of instructions and reinforcement-feedback on human operant behavior maintained by fixed interval
reinforcement. Journal of the Experimental Analysis of Behavior, 12, 701–712.
Baron, A., & Perone, M. (1982). The place of the human subject in the operant laboratory. The Behavior Analyst, 5, 143–158.
Baeyens, F., Eelen, P., & Van den Bergh, O. (1990). Contingency awareness in evaluative conditioning: A case for unaware affective–evaluative learning.
Cognition & Emotion, 4, 3–18.
Bentall, R. P., & Lowe, C. F. (1987). The role of verbal behavior in human learning. 3. Instructional effects in children. Journal of the Experimental Analysis of
Behavior, 47, 177–190.
Bentall, R. P., Lowe, C. F., & Beasty, A. (1985). The role of verbal behavior in human learning. 2. Developmental differences. Journal of the Experimental Analysis
of Behavior, 43, 165–181.
Author's personal copy

64 C.A. Bradshaw, P. Reed / Learning and Motivation 43 (2012) 55–65

Brewer, W. F. (1974). No convincing evidence for conditioning in adult humans. In W. B. Weimer, & D. S. Palermo (Eds.), Cognition and the symbolic processes
(pp. 1–42). Hillsdale, NJ: Erlbaum.
Catania, A. C., Lowe, C. F., & Horne, P. (1990). Nonverbal behavior correlated with the shaped verbal behavior of children. The Analysis of Verbal Behavior, 8,
43–55.
Catania, A. C., Matthews, B. A., & Shimoff, E. (1982). Instructed versus shaped verbal human behavior: Interactions with nonverbal responding. Journal of
the Experimental Analysis of Behavior, 38, 233–248.
Catania, A. C., Matthews, T. J., Silverman, P. J., & Yohalem, R. (1977). Yoked variable-ratio and variable-interval responding in pigeons. Journal of the
Experimental Analysis of Behavior, 28, 155–161.
Catania, A. C., & Reynolds, G. S. (1968). A quantitative analysis of the responding maintained by interval schedules of reinforcement. Journal of the Experimental
Analysis of Behavior, 11, 327–383.
Dack, C., McHugh, L., & Reed, P. (2010). The transfer of judgments of control to a discriminative stimulus and to novel stimuli through derived relations.
Learning and Behavior, 38, 348–366.
Dawson, M. E., & Schell, A. M. (1985). Information processing and human autonomic classical conditioning. In P. K. Ackles, J. R. Jennings, & M. Coles (Eds.),
Advances in psychophysiology. Greenwich, CT: JAI Press.
Dickinson, A., & Brown, K. J. (2007). Flavor-evaluative conditioning is unaffected by contingency knowledge during training with color-flavor compounds.
Learning & Behavior, 35, 36–42.
Ericsson, K. A., & Simon, H. A. (1980). Verbal reports as data. Psychological Review, 87, 215–251.
Ferster, C. B., & Skinner, B. F. (1957). Schedules of reinforcement. New York: Appleton-Century-Crofts.
Harzem, P., Lowe, C. F., & Bagshaw, M. (1978). Verbal control in human operant behavior. Psychological Record, 28, 405–423.
Hayes, S. C. (1989). Rule-governed behavior: Cognition, contingencies and instructional control. New York: Plenum Press.
Hayes, S. C., Brownstein, A. J., Zettle, R. D., Rosenfarb, I., & Korn, Z. (1986). Rule-governed behavior and sensitivity to changing consequences of responding.
Journal of the Experimental Analysis of Behavior, 45, 237–256.
Hayes, S. C., & Hayes, L. J. (1992). Verbal Relations and the Evolution of Behavior Analysis. American Psychologist, 47, 1383–1395.
Hofmann, W., De Houwer, J., Perugini, M., Baeyens, F., & Crombez, G. (2010). Evaluative conditioning in humans: A meta-analysis. Psychological Bulletin,
136, 390–421.
Kaufman, A., Baron, A., & Kopp, R. E. (1966). Some effects of instructions on human operant behavior. Psychonomic Monograph Supplements, 1, 243–250.
Laties, V. G., & Weiss, B. (1963). Effects of a concurrent task on fixed-interval responding in humans. Journal of the Experimental Analysis of Behavior, 6,
431–436.
Leander, J. D., Lippman, L. G., & Meyer, M. M. (1968). Fixed interval performance as related to subjects’ verbalizations of the reinforcement contingency. The
Psychological Record, 18, 469–474.
Lippman, L. G., & Meyer, M. M. (1967). Fixed interval performance as related to instructions and to subjects’ verbalization of the contingency. Psychonomic
Science, 8, 135–136.
Lowe, C. F. (1979). Determinants of human operant behavior. In M. D. Zeiler, & P. Harzem (Eds.), Advances in analysis of behaviour: Vol. 1. Reinforcement and
the organisation of behaviour (pp. 71–93). Chichester, UK: Wiley.
Lowe, C. F., Beasty, A., & Bentall, R. P. (1983). The role of verbal behavior in human learning: Infant performance on fixed interval schedules. Journal of the
Experimental Analysis of Behavior, 39, 157–164.
Lowe, C. F., Harzem, P., & Bagshaw, M. (1978). Species differences in temporal control of behavior. II. Human performance. Journal of the Experimental
Analysis of Behavior, 29, 351–361.
Lowe, C. F., Harzem, P., & Hughes, S. (1978). Determinants of operant behaviour in humans: Some differences from animals. Quarterly Journal of Experimental
Psychology, 30, 373–386.
Matthews, B. A., Catania, A. C., & Shimoff, E. (1985). Effects of uninstructed verbal behavior on nonverbal responding: Contingency description versus
performance descriptions. Journal of the Experimental Analysis of Behavior, 43, 155–164.
Matthews, B. A., Shimoff, E., Catania, A. C., & Sagvolden, T. (1977). Uninstructed human responding: Sensitivity to ratio and interval contingencies. Journal
of the Experimental Analysis of Behavior, 27, 453–467.
Mitchell, C. J., De Houwer, J., & Lovibond, P. F. (2009). The propositional nature of human associative learning. Behavioral and Brain Sciences, 32, 183–246.
Osborne, L. A., & Reed, P. (2008). Parents’ perceptions of communication with professionals during the diagnosis of autism. Autism, 12, 259–274.
Peele, D. B., Casey, J., & Silverberg, A. (1984). Primacy of interresponse-time reinforcement accounting for rates under variable-ratio and variable-interval
schedules. Journal of Experimental Psychology: Animal Behavior Processes, 10, 149–167.
Pleyers, G., Corneille, O., Luminet, O., & Yzerbyt, V. (2007). Aware and (dis)liking: Item-based analyses reveal that valence acquisition via evaluative
conditioning emerges only when there is contingency awareness. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33, 130–144.
Raia, C. P., Shillingford, S. A., Miller, H. L., & Baier, P. S. (2000). Interaction in procedural factors in human performance on yoked schedules. Journal of the
Experimental Analysis of Behavior, 74, 265–281.
Randell, J., Ranjith-Kumar, A. C., Gupta, P., & Reed, R. (2009). Effect of schizotypy on responding maintained by free-operant schedules of reinforcement.
Personality and Individual Differences, 47, 783–788.
Reed, P. (1993). Influence of schedule of outcome presentation on human judgments of causality. Quarterly Journal of Experimental Psychology, 46A, 327–345.
Reed, P. (1994). Influence of the cost of responding on human causality judgments. Memory & Cognition, 22, 243–248.
Reed, P. (1999). Effect of perceived cost on judgments regarding the efficacy of investment. Journal of Economic Psychology, 20, 657–676.
Reed, P. (2001). Schedules of reinforcement as determinants of human causality judgments and response rates. Journal of Experimental Psychology: Animal
Behavior Processes, 27, 187–195.
Reed, P. (2006). Effect of required response force on rats’ performance on a VI+ schedule of reinforcement. Learning and Behavior, 34, 379–386.
Reynolds, G. S. (1975). A primer of operant conditioning. Glenview, IL: Scott, Foreman.
Shanks, D. R., & Dickinson, A. (1991). Instrumental judgment and performance under variations in action-outcome contingency and contiguity. Memory
and Cognition, 19, 353–360.
Shimoff, E., Catania, A. C., & Matthews, B. A. (1981). Uninstructed human responding: Responsivity of low-rate performance to schedule contingencies.
Journal of the Experimental Analysis of Behavior, 36, 207–220.
Shimoff, E., Matthews, B. A., & Catania, A. C. (1986). Human operant performance: Sensitivity and pseudosensitivity to contingencies. Journal of the
Experimental Analysis of Behavior, 46, 149–157.
Skinner, B. F. (1966). An operant analysis of problem solving. In B. Kleinmuntz (Ed.), Problem-solving: Research, method, and theory (pp. 225–257). New York:
Wiley.
Skinner, B. F. (1969). Contingencies of reinforcement: A theoretical analysis. New York: Appleton-Century-Crofts.
Stahl, C., & Unkelbach, C. (2009). Evaluative learning with single versus multiple USs: The role of contingency awareness. Journal of Experimental Psychology:
Animal Behavior Processes, 35, 286–291.
Stahl, C., Unkelbach, C., & Corneille, O. (2009). On the respective contributions of awareness of US valence and US Identity in valence acquisition through
evaluative conditioning. Journal of Personality and Social Psychology, 97, 404–420.
Vaughan, M. (1989). Rule-governed behavior in behavior analysis. In S. C. Hayes (Ed.), Rule-governed behavior: Cognition contingencies, and instructional
control (pp. 97–118). New York: Plenum Press.
Vaughn, S., Schumm, J. S., & Sinagub, J. (1996). Focus group interviews in education and psychology. London: Sage.
Walther, E., & Nagengast, B. (2006). Evaluative conditioning and the awareness issue: Assessing contingency awareness with the four-picture recognition
test. Journal of Experimental Psychology: Animal Behavior Processes, 32, 454–459.
Author's personal copy

C.A. Bradshaw, P. Reed / Learning and Motivation 43 (2012) 55–65 65

Weiner, H. (1964). Conditioning history and human fixed-interval performance. Journal of the Experimental Analysis of Behavior, 7, 383–385.
Weiner, H. (1969). Controlling human fixed-interval performance. Journal of the Experimental Analysis of Behavior, 12, 349–373.
Weiner, H. (1970). Human behavioural persistence. Psychological Record, 20, 445–456.
Zuriff, G. E. (1970). A comparison of variable-ratio and variable-interval schedules of reinforcement. Journal of the Experimental Analysis of Behavior, 13,
369–374.

You might also like