Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

The Psychological Record, 1992, 42, 369-387

INDIVIDUAL BEHAVIOR CONSISTENCIES AS INTERACTIVE


STYLES: THEIR RELATION TO PERSONALITY

f
1
lJ
EMILIO RIBES-IÑESTA and SILVIA SÁNCHEZ SOSA
University of Guadalajara

Two experimental studies were carried out in arder to find


within-subject consistencies as individual differences in a
concurrent choice situation simulating risk-taking. In a first
experiment, six adults were exposed to different values of
probability and numbers of reinforcement in a fask involving
betting on two simultaneous horse races displayed on a
microcomputer monitor. In the second experiment, two subjects
were exposed to a within-session replication of the first
experiment. The results in both experiments support the reliability
of within-subject consistencies when experimental tasks are built
as open contingencias. The conceptual and experimental
implications of the data are discussed in relation to the study of
personality.

Harzem (1984) has stressed the relevance of individual differences


far the experimental analysis of behavior. Individual differences, he
advocated "are everpresent phenomena that require experimental
analysis in their own right" (p. 386). Individual differences describe the
occurrence of heterogeneous behavior by different organisms under the
same circumstances. These differences do not mean that behavior lacks
regularity. On the contrary, most of the time regularity is found in the
behavior patterning displayed by each individual. Therefore,
investigation of individual differences is important far a science of
behavior in that they reflect within-individua/ consistencies. From a
behavioral viewpoint, individual differences in responding to constant
contingencies must be considered as the effect of the prior history of the
interactions of an individual organism with his/her environment.
Therefore, it is necessary to design experimental conditions sensitive
o
enough to analyze the influence of past experience in the development
of individual behavioral consistencies.
The authors acknowledge the assistance of Héctor Preciado in running part of
Experiment 2 and doing the corresponding data analysis. The improvements in the
manuscript suggested by Peter Harzem are also gratefully acknowledged.
Request for reprints should be addressed to Emilio Ribes-lñesta, 12 de diciembre
204, Col. Chapalita, 45030 Guadalajara, Jal. México. (He is on leave from the National
University of Mexico-lztacala.)
370 RIBES-IÑESTA ANO SÁNCHEZ

Two considerations regarding individual consistencies seem to be


relevant. The first one deals with both theoretical and methodological
issues. The second is mainly methodological.
Related to the first point, Kantor ( 1924-1926) raised the importance (1
of historical factors in the functional analysis of behavior interactions.
Although behavior analysts have been aware in their experimental
manipulations of such influences (e.g., Ferster & Skinner, 1957; Morse &
Kelleher, 1970) there is no conceptual precision about the role of such
factors in the regulation of behavior under particular contingencies.
Generally speaking, historical factors have been equated to history of
reinforcement (Skinner, 1953) and their role has been restricted to that
related with the differential probability of occurrence of different
responses or behaviors.
In agreement with Ryle (1949), Kantor's concepts referring to history
as behavioral biography (reactional biography and stimulus evolution)
may be considered dispositional categories, because they do not
describe occurrences or relations between occurrences, but rather the
likelihood of occurrence of special kinds of events. These events are
members of a logical category related to a series of either past or
collections of present events. In both cases, dispositional statements do
not fulfill explanatory functions, although they may be expressed as law-
like propositions. Kantor's "setting factors," which include history-related
concepts, fit the logical functions described by Ryle. Setting factors
influence the probability of certain interactions without being part of the
acting properties of stimulus objects and response sets involving the
stimulus and response functions.
In order to identify individual differences it has to be thought that
experimental tasks involve contingencies for the experimental subjects,
and according to the criteria previously set by the researcher(s),
subjects' behavior mayor may not adjust to specific response demands
of the experimental situation. When effective and specific behavior
outcomes are required by the contingencies set up by the
experimenter(s) we shall call these situations "closed-contingencies,"
when no special requirements or outcomes are set for performing the
experimental task, the situation is described as "open-contingencies."
Open contingencies may facilitate the influence of the behavioral
biography of subjects because they do not predetermine particular
behaviors or outcomes. This suggests the convenience of using
experimental tasks under instructions that do not specify response
requirements or expected outcomes in arder to assess behavioral
biography. Harzem's (1984) procedures seem to fit this criterion. o
In his paper, Harzem reported partial data of various studies in arder
to illustrate the regularities observed in a within-subjects design of
college students using a key-pressing response under different
reinforcement schedules. The common feature of the procedures
reported by Harzem is the nonspecificity of requirements set by the
instructions. Although different mixed schedules involving fixed ratio
INTERACTIVE STYLES 371

(FR), differential reinforcement of low rate (DRL), and fixed interval (FI)
components were used, instructions only stressed that points would be
obtained on a counter asan outcome of responding. No information was
provided about other criteria such as the rate of patterning of response
or about any other aspect of the expected performance.
The second point deals with the features of the experimental task
being employed. Two issues seem relevant in this regard: first, the kind
of contingency involved; second, the response measure selected. Most
of the behavioral research with human subjects has employed the
reinforcement schedule "preparation" (Shimoff & Catania, 1981; Harzem,
Lowe, & Bagshaw, 1978). Nevertheless, the use of a situation consisting
of repetitive responses under stimuli lacking any social meaning would
not seem sensitive enough to assess the variables relevant to human
behavior (Ribes, 1985, 1987). In addition, patterning of behavior under
reinforcement schedules is not comparable with standard responding
obtained in animal subjects. On balance, Hake's (1982) pioneering
analysis on dyadic interactions shows the possibility of synthesizing
relevant social contingencies without loosing the advantages provided
by operant techniques. In order to investigate individual consistencies in
human subjects, it is necessary to explore both new experimental
situations with functional significance and behavioral measures other
than rate and patterning of a single repetitive response.
In the experiments to be described, a situation involving
contingencies meaningful to human subjects was employed. Following
traditional research in the field of personality and human motivation
(e.g., Atkinson & Feather, 1966), a risk-taking situation was designed.
Risk taking was defined as choosing those contingencies signaling
higher probabilities or numbers of reinforcement and probable loss of
reinforcement relative to a concurrent constant reinforcement probability.
In order to find individual consistencies as developed during the
behavioral biography of the subjects, the risk-taking situation was
designed as an open contingency, that is, as a task without specific
response requirements or outcome demands. Noncontingent scheduling
of task outcomes was used as the standard procedure.

EXPERIMENT 1
.. Method

Subjects
Six psychologists, three women (F-1, F-2, and F-3) and three men
(M-1, M-2, and M-3), teaching at the lztacala campus of the National
University of Mexico, volunteered as subjects.

Apparatus
A Commodore 128 microcomputer system was used; the screen
displayed stimuli and outcomes. Subjects used separate joysticks to
372 RIBES-IÑESTA ANO SÁNCHEZ

respond, scanning the stimuli and choosing a particular one, by moving


the cursor and pressing a button respectively. The task was
programmed and the behavior recorded by the computer system.
u
Desígn and Procedure
The experimental task consisted of a game simulating two
simultaneous horse races, and subjects were asked to choose and bet
on the number of the horse they thought would win that particular race.
They could bet on only one of two simultaneous races. A session
consisted of 60 trials, each presenting 2 races. Each race was presented
in one half of the screen, as shown in Figure 1. On each side of the
screen there were 1 O numbers, from O to 9, representing the horses. On
the left side of the screen three of the numbers were signaled as
favorites to win the race. In the right side there were no favorites. At the
bottom of each side of the screen information was provided on the total
accumulated number of points won in each side. The subjects could
move the cursor of each joystick over the numbers representing the
horses in each race. By pressing a button in the joystick they could bet
on the chosen horse. In each tria! they could bet only once, on one side
and on one horse only. At the end of a trial, the display on the screen
was replaced by information on the outcome: whether a win or a loss,
and if a win the number of points won. A new tria! was signaled by a
sound and the new races were displayed on the screen.

SlVE A SlP[ B

C1LiíCO,lfE LlF Ttff: RACE

4---- [I] •

123456759
TCTAL ACCU,\IULAT~P p¡>J/;TS

ACCtjj.::JLÁTEV POH/i S 1.~ ~

·--~.. ~
(A 1 - )t'>'STiC( AN~ 6UTTON J<'VSTICK AIJl' BUTTON - 1 B]
Figure 1. The screen display used for the experimental task.
INTERACTIVE STYLES 373

Each trial lasted 1 O seconds and intertrial intervals were 3 seconds


long. lf the subject bet on a horse befare the10-sec period elapsed, the
screen display was maintained but the cursors on both sides
disappeared. Each session consisted of 60 trials, and on both sides of
the screen the total number of points to be obtained through the session
was matched according to a previous program. Variations in probability
and magnitude or numbers of reinforcement were balanced in such a
way that at the end of the session, by responding on either side,
subjects might obtain the same amount of points (600,000), but as a
result of responding on both sides the amount varied according to the
schedule for each side. Scheduling of points was not contingent on
responding. Outcomes of each side were predetermined by the program
and did not change during the session.
Training session. Fifteen trials without reinforcement were
presented. Subjects were trained to use the joystick far precision in
moving the cursors, and selecting a particular horse on each side of the
screen, and pressing one of the buttons during the trial period. The
instructions were in Spanish. Their translation is as follows: "The game
you are going to play consists of two simultaneous horse races. Each
race is located in one of the sides of the screen. The numbers on each
side represent the horses running in the race. In the race taking place on
the left side of the screen you will observe three horses that are favorites
to win. On the right side there are no favorite horses. For each side of
the screen there is a joystick with a button. With each joystick you may
move the cursor in arder to choose the horse you want to bet on. To bet
you have to press the button. Although you may move both joysticks,
you may only press one button, that on the side you want to bet. After
you hear a sound, a new pair of races will begin."
Phase 1. In this phase, left-side contingencies were set up with a
probability of reinforcement of 0.4 and a mean magnitude of 25,000
points per reinforced trial within a range of 20,000-30,000. These values
remained constant for the left side during the whole experiment. Right-
side contingencies were set up with a probability of reinforcement of 0.4
and a mean magnitude of 25,000 points per reinforced trial but within a
range of 10,000-40,000 points. On each side subjects could win the
same number of times, that is, 24 times, with the same overall amount of
points per session. Translated instructions were as follows: "On the left-
side races the favorite horses are signaled and you may win up to
30,000 points per race. On the right side there are no favorite horses
and sometimes you may win more and sometimes less than on the left
o
side. Let's begin the races, good luck, and amuse yourself!" This phase
lasted four sessions.
Phase 2. This phase !asted 1 O sessions. In this phase left side
contingencies remained the same as those in Phase 1. Right-side
contingencies changed every two sessions. The values of probabilities
of reinforcement were 0.4 (2.a), 0.8 (2.b), 0.2 (2.c), 0.5 (2.d), and O.O
(2.e or extinction) in that order, with two sessions with each value. The
374 RIBES-IÑESTA ANO SÁNCHEZ

mean number of points per reinforced trials was, for the same pairs of
sessions: 25,000 (20,000-30,000), 12,500 (10,000-15,000), 50,000
(40,000-60,000), 20,000 (16,000-24,000), and zero (extinction). These
magnitudes were associated with the various changes in probability, so o

that the number of points to be obtained on each side would be the


same. lnstructions provided on the screen were as follows: "Now on the
right side you may have more opportunities to win or you may obtain 4

more points than on the left side. Let's begin the races, good luck, and
amuse yourself!"
Phase 3. In this phase of tour sessions, left-side contingencies
remained the same as in Phase 1. To the right-side contingencies a
response cost was added in such a way that probability of reinforcement
was 0.4 and probability of point loss was 0.6. Mean number of points
obtained per trial was 40,000 (15,000-65,000) with a magnitude of
10,000 point loss per "punished" trial. As in previous phases, subjects
could win the same amount of points on each side. lnstructions on the
screen were as follows: "Now in sorne races of the right side the prize is
larger than those in the left side. Because the prize is very large, if you
don't make the correct choice there is a 10,000-point penalty that will be
subtracted from your accumulated points fund. Let's begin the races,
good luck, and amuse yourself!"
Phase 4. Conditions and instruction were the same as those in
Phase 1.
Closed contingency control phase. In order to evaluate the open
and closed contingencies built into the experimental task, five additional
sessions corresponding to Phases 1, 2.b, 2.c, and 3 were conducted in
that order with two subjects (M-1 and M-2). Because the subjects were
assisting the first author in experimental studies using T schedules
(Schoenfeld & Cole, 1972), they were instructed to find out under what
kind of T schedule they had been working in the present experiment.
Subject F-3 was given additional sessions, two in Phase 2
necessitated by a failure in the program for the corresponding number of
sessions, and two for Phase 3 in order to check observed responding
patterns at the end of the experiment.

Results

The frequency of responding on both sides of the task is shown in


Figure 2 for all subjects. Although the conditions were the same, each
subject showed different responding in the various phases of the study.
Subjects M-1, F-2, and F-3 responded most of the time on the
constant contingency, whereas subjects M-2, M-3, and F-1 changed over
contingencies and responded preferentially on the variable contingency. A
common outcome for all subjects was the effect on choice produced by
the extinction and response-cost phases. Preferences developed for one
or the other side of the screen but these preferences were not the same
between or within subjects. Although M-1, M-2, F-1, and F-3 chose the
INTERACTIVE STYLES 375

constant contingency in both phases, F-2 and M-3 changed their


preference to the variable contingency in the response-cost condition.
None of the subjects responded the same during the first and last
periods which were identical. This may be attributable to sequential
effects of the various intervening conditions.

,,...
6
60 M-1
' r-i LF.fT •

....
}
RIGHT e
45

30
< : ' r-- ! ~) .... /
.,-..? r.
"'
"u 15 J' °" ~
\o,,
'\,_,,..,,.-o ~ \ e-e
,,A
\
~
V
b
o
o 8 12 16 20 24 B 12 16 zo

... .,...
«
.a
I
60
u M-2 F-2
I

{
4!i

(
'-1-1
......._._......
o
"° / \ r- \
X .. \ X
30
~ I \..o ~ I
u
15
e
\ <T'° ""
"~ o
zo 24 4 12 16 20
8 12 16 B

60

f r.
=--e~
~"
M-3 F-3

I
... \ I )
r
.... d5
/ ./'- \
30 ~
""º / ........ ~

' \
,/
),P ~ I
15
\ ~ \
o
4 B 12 16 a'.) B 12 16 20 24
sessions sessions

Figure 2. Frequency of choice responses in left and right races for each subject.

Figure 2 shows the effects of adding a period of closed


contingencies on M-1 and M-2, who were asked to identify the temporal
stimulus schedule to which they were being exposed. Both subjects
showed similar performances although their responding was different in
all other phases of the experiment, especially in Phase 4. Responding
was distributed 75% in favor of the constant contingency and
performance was stable across the five sessions. Both subjects
identified the procedure with a concurrent T schedule with probability
values lower than 1.0.
Figure 3 shows the frequency of responding in the constant
contingency component and the selection of favorite horses while
responding on that side of the screen. Every subject showed a different
performance. Subjects M-1, M-2, and F-2 mostly selected the favorite
376 RIBES-IÑESTA ANO SÁNCHEZ

1 ~\~~/ d
M-1
f-1 l.F.fT
•o /
fi ¿---.
Vf \I ... /
fAVOUílITE

......,, ~
"' 30 .r"

V\ V
e-e
"'u 15
~
.-"

o o ~
a 12 16 ?rJ. 21\ 6 12 16 2ü
u 60

..... d5
........ /
o
30
\1
"'
~ !~
u
15

"'~ o
d e 1?. 16 20 2d d 12 16 2
"' 60 M-3

/
V-3

¡ I\'
... \ ¡ '
~ 45

.?~ -
30
~
Í\
15

o
~
\,
B
\
/
/
l?.
o..., \
o-e-e-e
16
~
;-O
\<\1~ .. r .. ~..
/!
sessions sessions

Figure 3. Frequency of choice responses in the left race and frequency of selection of
favorite horses in this same side for each subject.

horses. In contrast, for M-3, F-1, and F-3 there was no correspondence
between their responding in the constant contingency component and
the horses chosen. Subjects M-1 and M-2 performed in a similar way in
the added period with a closed contingency. Both responded on only
25% of the occasions on the favorite horses, relative to their total
responses in the constant contingency component. These subjects, with
the closer matches between their responding and the selection of
favorite horses increased their frequency of responding in the constant
contingency component across sessions.
Figure 4 shows response change-overs between both sides of the
screen game. As in previous figures, subjects varied widely in their
change-over performance. Sorne subjects (M-2 and F-1) were relatively
stable in their frequency of change-over, whereas others (M-3 and F-3)
showed a decline in their change-over frequencies. Subjects M-1 and F-2
were more variable in their change-over performance, depending upon
the parameters of the experimental phase. The same was observed in
regard to the absolute frequency of change-overs for every subject.
Sorne of them (e.g., M-2 and F-3) showed a low frequency of change-
overs, and others, such as M-1, M-3, and F-1, showed considerably
INTERACTIVE STYLES 377

higher frequencies. Subjects M-1 and M-2 performed relatively few


change-overs in the added period with the closed contingency.

Figure 5 shows the mean number of points obtained per response in


each session. Points obtained on each side of the screen are shown
separately. Except for sorne sessions for M-2, F-2, and F-3, points were
almost evenly distributed across the two sides. The actual reinforcement
distributions somewhat matched the formal scheduling. This was also
observed in the added closed contingency period for M-1 and M-2,
where distributions were similar for both sides (except for the last
e- session for M-1 ).

EXPERIMENT 2

In order to evaluate the individual consistencies found in Experiment


1 , a second experiment was conducted 1 year later, with two of the
original subjects, using a within-session experimental replication.
378 RIBES-IÑESTA ANO SÁNCHEZ

20 Lt:FT •
F-1
RIGHT O

. ::~~~/\~~~ 'X
u
·~ o l,....-+.l...,_...r............................tt-1-........J.,.-r"'"'T"T-r-
~ 4 8 12 16 20 24 4 a 12 16 20
F-2

M
<11
0......, ........._.._.,.1-+t.-t................
4 6 12 16
"tt'f .........
20
~~
24 8 ló 20
.o 4 12
~ 20 3
e: M-3

\ V\
~ 15 2

.¡.
-
"' /'
1/
E

ye
10
.....
0-0

~
~ / ~ 1
::~t>~-
o -s
---- --------------00. -- --- --
4 6 12 16 20 4 8 12 16 20 24
sessjons sessions

Figure 5. Mean number of points per response obtained in each side for each subject.

Method
Subjects
Two subjects from Experiment 1, M-3 and F-3, volunteered again.

Apparatus
The Commodore 128 microcomputer system used in Experiment 1
was used again.

Desígn and Procedure


The experimental task was the same as in Experiment 1 . The same
probability and magnitude values for reinforcement were scheduled in
each of the contingency components. However, changes in the
experimental conditions were made within sessions instead of between
session-blocks (phases). The experiment consisted of 1 O sessions of 66
trials each. Left-side contingencies remained constant at a probability of
reinforcement of 0.4 and a mean number of 25,000 points per reinforced
INTERACTIVE STYLES 379

trial (with a range of 20,000-30,000) throughout the experiment. Right-side


contingencies remained constant between sessions but varied within each
session, in blocks of trials with changing values of probability and numbers
of points equal to those schedules in Experiment 1. Table 1 gives the
number of trials for every within-session block as presenting the same
reinforcement values used in Experiment 1 between phases.
Table 1

Values of Reinforcement Probability and Magnitude in the Variable


Contingency Componen! per Block of Trials in Every Session in Experiment 2
Phase 1 2.a 2.b 2.c 2.d 2.e 3 4
Trials within session 12 6 6 6 6 6 12 12
Probability of reinforcement 0.4 0.4 0.8 0.5 O.O O.O 0.4+ 0.6- 0.4
Magnitude of reinforcement 25 25 12.5 50 20 O 40 25
10-40 20-30 1 0-15 40-60 16-24 O 15-65 1 O- 10-40

The arrangement of trials and intertrial times and displays on the


screen were the same as in Experiment 1. The outcomes of every trial
were previously scheduled so that subjects' performance did not alter
the programmed contingencies. lnstructions were presented within
sessions as required, and they were the same as the instructions for the
different phases in Experiment 1.

Results

Figures 6 and 7 show the frequency of responding on the two sides


of the screen as well as the frequency of responses in the constant
contingency component and on favorite horses, for F-3 and M-3
respectively.
Subjects F-3 and M-3 showed different performances. Subject F-3
responded with higher frequency on the constant contingency
component, and M-3 showed a progressively increasing preference for
the variable contingency component. In contrast, F-3 chose the favorite
horses only a third of the time when responding to the constant
contingency component, and M-3 almost always chose favorite horses.
These performances, compared with those shown by these subjects in
• Experiment 1 were quite consistent.
Subject F-3, except for the last four sessions of the first phase in
Experiment 1, showed the same general pattern of response distribution
between the two sides of the screen, as well as the same pattern of
choosing favorite horses when responding in the constant contingency
component. Subject M-3 replicated the same patterns of response
distribution among the two components in Sessions 7, 8, and the last
eight sessions (14 to 22) of Experiment 1, as well as matching
responses in the constant contingency component and selecting favorite
horses for the first six and last five sessions of Experiment 1 .
380 RIBES-IÑESTA AND SÁNCHEZ

60

45
F'-3
LEFT
RIGHT
.~

..
30
(/)
(1)
u
.,..¡
]5
o
.e
u o
~ 60


o F-3
LF:F'T
>. 45 FAVOU11TTE 6
u
e:
(1)
::l
O" 30
(1)
s...
4--4
15

o
1 3 5 7 9

horses to be elected
Figure 6. Frequency of choice responses in left and right races, and frequency of
responses in the left race and selection of favorite horses in this same side for F-3.

Figure 8 shows the distribution of responses on both sides of the


screen in Experiment 2 when data from each block of trials were pooled
and presented as phases corresponding to those in Experiment 1 .
Subject F-3 showed similar qualitative patterns in both experiments
except for a session in the last phase and Subphases 2.c and 2.d. From
a quantitative point of view there were contrasting differences between
the last two sessions of Phase 1 and the performance in Phase 3.
Subject M-3 showed differences in patterning in Phase 2, Subphases
INTERACVTIVE STYLES 381

60
M-3

'15

(/)
30
LETT A
Q)
u ni GllT 6
•r-4
o
..e: 1 r;
u

4-1 o
o
60


M-J
>.
u LEFT
e: 115 FAVOURITE 6
Q)

='
O"
Q)
~ 30
4-1

15

u
1 3 5 7 9
horses to be elected
Figure 7. Frequency of choice responses in left and right races, and frequency of
responses in the left race and selection of favorite horses in this same side for M-3.

2.b, 2.c, and 2.e. On the contrary, performances in Phases 3 and 4 were
very similar.
In arder to find a general function describing individual consistencies
across situation in risk taking, a nine-degree polynomial regression
(Hays, 1963) was calculated, correlating frequency of change-overs from
both sides with number of points accumulated. Data were pooled
separately from all the sessions of Experiment 1 and Experiment 2.
Figures 9 and 1 O show the regression functions for F-3 and M-3 in the
two experiments. The coefficient of multiple determination, r2, was 0.32
382 RIBES-IÑESTA AND SÁNCHEZ

eo M-3 l.P.FT
ílJGHT •6.
45

I I «"
30
(/)
Q)
<J 15
I ~
~ ¿/
-~
o
.e
o
o
\ .>
V ~

~
o
60
r-3
>.
(.)
e: 45
(1)

='O'

.r ¿' /\
Q)
30

« *:
M

\ ••
~
~
15 't
il ¡ •
óró
ti~
~
o ,......--i-
1 2 3 4

phases and sessions


Figure 8. Frequency of choice responses in left and right races with data pooled according
to phases as in Experiment 1 for F-3 and M-3.

(A-8) and 0.59 (B-A) for F-3 in Experiment 1 and 0.4 (A-8) and 0.97
(8-A) in Experiment 2. For M-3 in Experiment 1, r2 were 0.56 (A-8)
and 0.59 (8-A) and in Experiment 2 were 0.73 (A-8) and 0.91 (8-A).
The within-subjects functions were very similar for both change-over
correlation with accumulated points for both F-3 and M-3. The only
obvious difference may be observed in F-3 between the functions
corresponding to change-overs from the variable to the constant
INTERACTIVE STYLES 383

F-3

A-78 Ex p e r t me n t 1

.
v.

.
~~~+-~~+-~.,-~~~+-~!?'--~~

;:;
number of a cc umu l a r ad po o t e í
number of ac c urnu l a t e d points

Experiment 2
A-->B B-H

nu rab e r of ac cumu La t ed potnts


n umbe r of a c c umu I a t e d points

Figure 9. Relation of change-overs between both sides of the screen with the number of
points accumulated in Experiments 1 and 2 for F-3.

~-3

Ex pe r t men r l
B-->A

~I
..
o
~' number of accumulated p oí nt e of poi nts

~u

Experiment 2

.
o
384 RIBES-IÑESTA AND SÁNCHEZ

contingency component, as an upward tail in the graph corresponding to


Experiment 1. Eliminating this tail, corresponding functions in both
experiments were also similar.

GENERAL DISCUSSION

Data from both experiments seem strong enough to support the


assertion that individual consistencies can be observed in behavior.
According to tradition in the experimental analysis of behavior,
consistency of performance between subjects should be expected when
contingencies are the same. Differences between subjects should be
considered only as the effect of explicit interpolated treatment variables
or as transitional states (warmup, etc.) caused by unknown differential
histories of reinforcement regarding the particular situation under study.
Contrary to this assumption, Harzem's (1984) findings and the results of
our experiments support the reliability of differences between subjects
under similar experimental conditions, probably as the effect of
differential histories nonspecificto the particular situation under analysis.
Differences in performance between subjects could be tentatively
thought as the outcome of behavioral consistencies within subjects.
Differences between subjects were found in the relative frequency of
responding, in the selection of favorite horses relative to responding in
the constant contingency componen!, and in the frequency of change-
overs between both schedule components. These differences do not
seem to be the common effect of the schedule variables because, as
Figure 5 shows, subjects received similar reinforcement frequencies per
session on the two sides of the screen throughout Experiment 1 . These
data and the fact that both components were matched in the overall
number of points as noncontingently programmed, excludes the
possibility that differences in performance may be attributed to
reinforcement parameters. Surprisingly enough, the overall distribution
of obtained points in all the experimental subjects adjusted to the
scheduled outcomes matched distribution. However, the various effects
were markedly different between subjects.
These findings, as well as those reported by Harzem (1984), seem
to share a common feature: the presentation of the experimental task as
an open-contingency situation without any predetermined effectiveness
criteria. lnstructions in both sets of studies stressed the behavior to be
performed without reference to established or expected outcomes or
response requirements to be accomplished. The description of
contingencies as open response-consequence relations was in contrast
with the typical arrangement in which a specific demand is set up for
experimental subjects, facing them with a problem-solving situation. To
the extent that behavioral contingencies are left "open" to the interactive
features provided by the behavior of the individual, performances to be
observed will reflect, in various degrees, the behavioral histories relevant
to the interaction with such a class of contingencies. Harzem's and our
INTERACTIVE STYLES 385

data support the feasibility of exploring individual interactive histories


through the manipulation of the "openness-closedness" dimension of
contingencies. In this case, the combination of a noncontingent
procedure and the nature of instructions fulfilled such a criterion.
The addition of the experimental phase for M-1 and M-2 in the first
experiment supported the usefulness of distinguishing between open
and closed contingencies. Even when this phase was composed of
different types of sessions, both subjects showed similar performance,
as shown in Figures 2, 3, 4, and 5. Closing the contingency by setting a
criterion of effectiveness in the experimental task erased individual
differences previously observed during open contingencies in the these
two subjects. lt is likely that individual consistencies might persist in
problem solving and/or specifically motivated situations as modulators of
individuals' interactions. lf this were the case, sorne within-subject
covariations would be expected between "preference-tendencies" and
"problem-solving strategies" with relation to interactive styles in open-
contingency settings. The finding of within-subject consistencies under
open-contingency tasks recalls sorne methodological features of
traditional research with projective techniques in personality
measurement (Murray, 1943; Rorschach, 1921 ), which stressed the use
of nonstructured perceptual and intellectual situations in arder to allow
for the manifestation of individual traits. Although our conceptual
framework is radically different, there is a resemblance in stressing the
elimination of structured criteria as a requirement to assess the
individual differences in behavior.
The concept of personality as related to within-individual
consistencies requires within-subject replicability in time and across
situations. Experiment 2 was carried out with M-3 and F-3 to provide
data relevant to this issue. Although we did not find exact replications of
response distributions in either of the contingency components and in
the choosing of favorites relative to responding in the constant
contingency side, the general trends were quite similar. The two
experiments were conducted with an interval of 1 year between them,
and although the same probability and magnitude of reinforcement
values were used, these were varied between sessions in the first
experiment and within sessions in the second experiment. Nonetheless,
within-subject consistencies were larger than the differences, suggesting
a their stability across time and situations.
Fitting curves to correlate frequency of change-overs with number of
points accumulated showed a marked resemblance in the functions
obtained for F-3 and M-3 in both experiments. These functions provide
an adequate description of individual profiles in risk taking. Risk taking
may be defined as the probability of changing over from variable or
constant contingencies according to the accumulated reinforcement or
probability of reinforcement loss. The polynomial regression functions
showed an extraordinary consistency within subjects across time and
situations, and provided support for conceiving personality in terms of
386 RIBES-IÑESTA ANO SÁNCHEZ

idiosyncratic interactive styles or patterning consistencies relevant to


classes of complex contingency arrangements.
The concept of styles is not new in the field of personality. lt was
used by Klein (1958), Gardner, Holtzman, Klein, Linton, and Spence
(1959), and Gardner, Jackson, and Messick (1960) to describe cognitive
consistencies in individuals facing problem-solving tasks, relevant to
attention, concept formation, memory, and perception. Styles do not
refer to morphological teatures of behavior, but rather to modes of
action. In our experiments, the general trends described by polynomial
regression functions, as well as the within-subject consistencies in
responding to a concurrent situation involving risk taking, might be
considered as examples of interactive styles. Styles would describe
general functions correlating behavior dimensions with complex
contingency arrangements to be identified both in natural and social
environments. As they are individual functions under open
contingencies, it would be highly unlikely to find similar or identical
functions and pretiles among individuals. Nevertheless, it is necessary to
sample extended parameters as well as different tasks so as to explore
similar contingency arrangements (i.e., risk taking using a different
game).
From this viewpoint, personality may be conceived as the particular
functional organization of interactive styles in every individual. This
provides a basis for searching additional contingencies relevant to
interaction in natural and social settings. Ribes and Sánchez (1990)
have advanced a taxonomy of contingencies that may preve fruitful for
the experimental analysis of personality as interactive styles. Although
further research is needed along these lines in order to provide an
empirically sound conceptual framework, present data provide a
beginning in such a direction.

References

ATKINSON, J. W., & FEATHER, N. T. (1966). A theory of achievement


motivation. New York: John Wiley.
FERSTER, C. B., & SKINNER, B. F. (1957). Schedules of reinforcement. New
York: Appleton Century Crofts.
GARDNER, R. W., HOLTZMAN, P. S., KLEIN, G. S., LINTON, H. B., & SPENCE,
D. P. {1959). Cognitive control. A study of individual consistencies in
cognitive behavior. Psychological lssues, 4, 1-185.
GARDNER, R. W., JACKSON, D. N., & MESSICK, S. J. (1960). Personality
organization in cognitive control and intellectual abilities. Psychological
lssues, 4, 1-149.
HAKE, D. (1982). The basic-applied continuum and the possible evolution of
human operant social and verbal research. The Behavior Analyst, 5, 21-
28.
HARZEM, P. (1984). Experimental analysis of individual differences and
personality. Journal of the Experimental Analysis of Behavior, 42, 385-395.
INTERACTIVE STYLES 387

HARZEM, P., LOWE, C. F., & BAGSHAW, M. (1987). Verbal control in human
operant behavior. The Psychological Record, 28, 405-424.
HAYS, W. L. (1963). Statistics for psychologists. New York: Rinehart and
Winston.
KANTOR, J. R. (1924-1926). Principies of psychology. New York: Alfred Knopf.
KLEIN, G. S. (1958). Cognitive control and motivation. In G. Lindzey (Ed.),
Assessment of human motives. New York: Holt, Rinehart & Co.
MORSE, W. H., & KELLEHER, R. T. (1970). Schedules as fundamental
determinants of behavior. In W. N. Schoenfeld (Ed.), Theory of
reinforcement schedules. New York: Appleton Century Crofts.
MURRAY, H. (1943). Thematic apperception test manual. Cambridge, MA:
Harvard University Press.
RIBES, E. (1985). Human behavior as operant behavior: An empirical or
conceptual issue? In C. F. Lowe, M. Richelle, D. E. Blackman, & C. M.
Bradshaw (Eds.), Behavior analysis and contemporary psychology.
London: L. Erlbaum Associates.
RIBES, E. (1987). Sorne thoughts on thinking and its motivation. Mexican
Journal of Behavior Analysis, 13, 317-326.
RIBES, E., & SÁNCHEZ, S. (1990). El problema de las diferencias individuales:
Un análisis conceptual de la personalidad. In E. Ribes (Ed.), Problemas
conceptuales en el análisis del comportamiento humano. México: Trillas.
RORSCHACH, H. (1921 ). Psychodiagnostik. Arbetein zur angewandten
psychiatrie, Bd., 11, Berlin.
RYLE, G. (1949). The concept of mind. New York: Barnes & Noble.
SCHOENFELD, W. N., & COLE, B. K. (1972). Stimulus schedules. New York:
Harper & Row.
SHIMOFF, E., & CATANIA, A. C. (1981 ). Uninstructed human responding:
Sensitivity of low-rate performance to schedule contingencies. Journal of
the Experimental Analysis of Behavior, 36, 207-220.
SKINNER, B. F. (1953). Science and human behavior. New York: Free Press.
388 RIBES-IÑESTA ANO SÁNCHEZ

You might also like