A History Of Rat Preference For Signalled Shock: From Paradox To

advertisement
Back to Realism Applied to
Home Page
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
History_Rat79.doc
A HISTORY OF RAT PREFERENCE FOR SIGNALLED
SHOCK: FROM PARADOX TO PARADIGM
GERLAD B. BIEDERMAN and JOHN J. FUREDY
University of Toronto
The evidential and logical status of preference-for-signal in the inescapable unmodifiable shock
context in rats was examined over a period that has seen the notion of a preference for signalled
shock develop from an apparent paradox in the existing conceptual framework of the 1950s to a
readily acceptable, established paradigm in the 1970s. The transformation of the phenomenon from
paradox to paradigm was found to be unwarranted from an assessment of the merits of symmetrical
and asymmetrical preference procedures upon which it was based.
The belief that there is a preference in humans
and animal subjects for signalled over
unsignalled inescapable and physically
unmodifiable noxious events, articulates so
well with most currently important theories of
behaviour (e.g., Berlyne, i960; Perkins, 1968;
Seligman, 1975), that it might appear foolish
to question the methodological and empirical
basis for this belief. In humans the preference
for signalling has recently been questioned in
a review of studies involving more than 570
subjects (Furedy, 1975), but it can be argued
that the issue is more critically resolved with
animal subjects, for whom the noxious event
can ethically be made genuinely aversive. In
this paper we direct our attention to a body of
literature with rats as subjects and shock as
the noxious, genuinely aversive event, where
it is clear that the general acceptance of
preference-for-signalling
notion
is
extremely
The research by the authors was supported by grants
from the National Research Council of Canada, Canada
Council, The Atkinson Charitable Foundation, Medical
Research Council of Canada, and the Health Sciences
Committee of the University of Toronto. We are grateful
to Professor W. O’Neill, Université de Moncton,
Moncton, New Brunswick, Canada, whose research on
the avoidance conditioning literature influenced some of
the historical analyses in this paper, and to J. M. Arabian
for comments on an earlier draft.
Requests for reprints should be sent to G. B.
Biederman, Scarborough College, University of
Toronto, West Hill, Ontario MIC IA4, Canada.
widespread.
Reviewers such as Seligman (1975) while
qualifying conclusions about a preference-forsignalled shock (PSS) phenomenon in humans
by citing Averill and Rosenn (1972) and
Furedy and Doob (1971, 1972) “for contrary
results”, make no hint of the necessity of any
such qualification in rats (Seligman, 1975,p.
199). D’Amato (1974) in another review of
the literature speaks of “a large number of
studies... which clearly show that animals ...
very generally prefer signaled to unsignaled
reinforcement”, and that the “experiments,
particularly those dealing with aversive
stimulation, seem to offer striking support for
the information hypothesis” (p. 93).
Experimenters also accept the existence of
PSS as the natural starting point for research.
For example, Cantor and LoLordo (1972)
state that “When an organism is permitted to
choose between a Pavlovian reinforcer
preceded by a warning signal and the same
rein-forcer unsignaled, he chooses the signaled reinforcer... with electric shock as the
reinforcer” (p. 259). MacDonald and Baron
(1973) claim that “the present results, together
with previous findings that animals choose
signaled shock in preference to unsignaled
shock support the hypothesis that signaling
shock reduces its aversiveness” (p. 37). In yet
another experimental report Arabian and
Desiderato (1975) state that
102
”Data from both human [and] animal ...
subjects have shown that, given the choice,
organisms will act to produce stimuli that
signal the imminence of a painful event” (p.
191). Finally, Miller, Daniel, and Berk (1974)
reaffirm that “rats ... are commonly believed
to prefer signaled shock to unsignaled shock
even when the shock is not modifiable” (p.
271).
Nevertheless, we shall suggest here that the
rat-PSS phenomenon is, in fact, a notion for
which the evidential support is equivocal. We
shall trace the history of PSS in rats as it has
been transformed from a paradox — from the
standpoint of a 1947 paper by Mowrer —
carefully investigated and cautiously
interpreted, into a paradigm, taken for
granted, and apparently needing no usual
rigorous validating checks, but only
demonstrations.
In what follows we shall outline the
evidential development of experiments after
1959 which have had a bearing on rat-PSS.
We are aware that any interpretation of
historical development in science may be
subject to reinterpretation; however, we wish
to emphasize that the methodological analysis
which forms a part of the historical account is
not subject to interpretational ambiguity as a
function of historical perspective.
This review will consider only literature
which used signal-contingent behaviour as the
dependent variable. Such other dependent
variables as conditioned emotional response
(CER) (e.g., MacDonald & Baron, 1973) and
ulceration (e.g., Weiss, 1970) have been cited
as relevant for PSS. However, even if one
assumes that these indices are valid measures
of “aversiveness” (MacDonald & Baron,
1973), it is still the case that the concepts of
aversiveness and preference, though related,
are different (cf. Furedy & Doob, 1971).
Thus, the present analysis will be limited to
studies which purport to be direct measures of
preference for signalled shock.1 All measures
of preference for signal must use signalcontingent behaviour but, as we hope to show
by analysis of a specific example, all signalcontingent behaviour-based measures are not
valid measures of preference. In particular,
we will propose that the only valid studies of
PSS are those which employ equivalent or
“symmetrical” behaviours to index both PSS
and preference for unsignalled shock (PUS).
THE CAUTIOUS, INVESTIGATORY
‘FIFTIES
Knapp, Kause, and Perkins (1959)
employed a symmetrical choice paradigm to
look specifically at preference in the T-maze.
These workers were quite open to both the
negative (conditioned-fear) and positive
(informational), or, in their terms,
“preparatory-response” aspects of the signal.
Their awareness of the negative aspects of
signalling is illustrated by their conclusion
that the “results of Exp. I could be interpreted
as supporting the hypothesis that S will make
a response followed by shock if this response
also eliminates a danger signal (a stimulus
that has regularly been followed by shock)
and the accompanying anxiety” (Knapp et al.,
1959, p. 361). On the other hand, they note
the positive aspects of signalling when they
say that the “results of Exp. II are not in line
with such an interpretation, as in this
experiment & acquired the response which
was followed by a danger signal (lights and
buzzer) rather than one which was not” (p.
361).
1
It may well happen that the ulceration and CER data
form the corpus of the only clear evidence for the
beneficial effects of signalling (e.g., Desiderato &
Newman, 1971; Hymowitz, 1976; Macdonald & Baron,
1973; Seligman, 1975; Weiss, 1970, 1971a, 1971b).
PSS as discussed in this paper is, as will be apparent, so
controversial, complex and perhaps inconsistent, that to
attempt to fully describe and analyse CER and
ulceration evidence in addition, is to wander too far
from our path and to attempt too long a trip.
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
103
Kamin (1956, 1957a, 1957b) which elegantly
investigated the prediction that WS
termination, despite the absence of US
termination, should lead to the learning of an
avoidance response (i.e., an escape-from-WSresponse). Kamin (1956) found that either WS
escape or US avoidance could equally well
serve as the reinforcing event for shuttling
behaviour. Other studies have also supported
the analysis of the importance of WStermination in the acquisition of avoidance
learning (Brush, Brush, & Solomon, 1955;
Church & Solomon, 1956; Verhave, 1959).
Other workers who have not accepted
Mowrer’s fear-conditioning analysis have
nevertheless used WS as a conditioned
aversive
reinforcer
whose
contingent
termination may increase some responses.
Dinsmoor (1954), Schoenfeld (1950), and
Sidman (1953) did not accept the notion of an
underlying fear-state, but still treated WS
termination as a reinforcing event.
By 1959, therefore, most psychologists, if
asked, would probably have not predicted
PSS. Indeed, on the basis of the conditioned
fear properties attaching to the WS, or on the
basis of the signal possibly causing one to
“tense up” (in layman terms), both the psychologists of the ‘fifties and the man in the
street would have had a reasonable
framework for predicting a preference for
unsignalled shock (PUS). The signal in the
inescapable shock context was therefore seen
as
having
predominantly
aversive
consequences.
THE ENTHUSIASTIC DEMONSTRATIONAL
SIXTIES
One paper which seems to have had a
significant role in converting caution
regarding PSS, as found in Knapp et al.
(1959), to enthusiasm is the often-cited report
by Lockard (1963). However, the report itself
was neutral with respect to predictions of
PSS; the purpose was to “test whether a
warning signal followed by unavoidable
shock
becomes aversive” (Lockard, 1963, p. 526).
The preference for the signalled side in this
shuttle-box symmetrical-choice experiment
was clear, reaching 90% in an orderly,
negatively accelerated acquisition function
over 12 days (Lockard, 1963, Figs 1 and 2),
so that it is perhaps not surprising that
reviewers became enthusiastic over the
informational, rewarding properties of signals,
and started to regard PSS as a phenomenon to
be demonstrated, rather than one whose
existence, because of the possibility of signals
becoming “aversive” (Lockard, 1963, p. 526),
was
still
an open question. The
demonstrational approach was probably
strengthened by the report of a check for the
scrambling factor where unauthorized
avoidances were monitored and found to be
both of low frequency of occurrence and
uncorrelated with PSS (Lockard, 1963, p.
529).
This way of checking on the scrambling
factor seems, in general, less effective in
comparison to the methods of Badia (Note 1),
and later, of Biederman and Furedy (1973),
who all manipulated scrambling and found
that PSS emerged only with unscrambled
shocks. Moreover, the Lockard method
measured unauthorized modifications on a
two-point scale (present or absent); later
studies by Biederman and Furedy (1973) and
Furedy and Biederman (1976) confirmed
Lockard’s
low
complete-modificationpercentage finding, but found that when
partial modification was measured by recording amount of shock received on each
unscrambled shock trial, there was a
significant relationship between amount of
unauthorized modification and amount of PSS
in the unscrambled-shock conditions.
Logically, of course, this scrambling
confound is not affected by the strength and
clarity of the PSS with unscrambled shock
reported by Lockard. However, there is a less
cited paper by the same author (Lockard,
1965) which repre-
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
104
sented a more extensive set of groups than
those in the earlier paper. The main finding of
the Lockard (1965) experiment is a weaker
PSS effect.2
In addition to Lockard’s (1963) paper the
other main evidential basis of the ‘sixties is
the paper by Perkins et al. (1966). In
discussing their five reported experiments
designed to investigate various factors behind
the PSS, Perkins et al. indicate their
confidence in the generality of PSS when they
note that there “does not appear to be a single
report of results indicating that unexpected
shock is less aversive than shock preceded by
a signal” (Perkins et al., 1966, p. 195), and
cite, the several previous studies which have
found PSS, (e.g., Perkins, Levis, & Seymann,
1963). This ready acceptance of PSS is based
on studies that all used unscrambled shock,
and- it will be recalled that the importance of
this factor had been empirically demonstrated
by Badia’s (Note 1) early study (Note 2).
The Perkins et al. (1966) study is
noteworthy also because it is often cited as
having overcome the modification problem
by using fixed electrodes in its last (fifth)
experiment presumably to overcome the
“postural” strategies mentioned by the more
cautious Knapp et al. (1959) paper, and the
more specific grid-shock-delivery problem
arising out of the failure to scramble the grids.
However, as Perkins et al. (1966, p. 194)
themselves indicate, so large a proportion of
subjects seemed to be able to detach the
electrodes during the course of the
experiment that the adjective “fixed” cannot
be justified in this procedure.
Nevertheless, this era in history of animal
PSS may be clearly charac2
It is interesting to note that a preference for
unsignalled shock (about 76%) emerges in Lockard’s
(1965, Fig. 1) paper, but signalled shock was correlated
with higher intensity shock (about 7%) making the PUS
finding not unequivocal.
terized as being “enthusiastic” in its general
acceptance of the phenomenon. As late as
1972, there is an empirical report (humansubject), stating that a “number of studies
show that... rats (e.g., Perkins et al., 1966) ...
consistently selected signaled over unsignaled
shock in a choice situation” (Suboski, Brace,
Jarrold, Teller, & Dieter, 1972, p. 407).
However, at least for such symmetrical
choice situations (generally based on the
shuttle-box apparatus), evidence of various
forms began to create considerable
difficulties for the PSS paradigm after 1966,
as detailed in the next section.
EVIDENTIAL PROBLEMS FOR THE PSS
PARADIGM IN SYMMETRICAL CHOICE
DESIGNS
The period of 1966 to the present can be
characterized as one which offers no
unequivocal support for strong animal PSS
based on statistically evaluatable criteria in an
unmodifiable-shock context. Moreover, there
are a number of published and publicallyreported instances where either no PSS or a
reliable PUS has been found. In other words,
the data-base is most consistent with Badia’s
(Note 1) early findings reported in a period
where animal PSS was more of a paradox
than a paradigm. This characterization of the
post-1966 evidence will now be documented
briefly.
The first class of evidence is that of journalpublished null findings, as represented by the
groups run with scrambled shock in the
experiments of Biederman and Furedy (1973)
and Furedy and Biederman (1976). The
second class of evidence is that of unpublished or non-journal-published null
outcomes. For obvious reasons the extent of
this class is impossible to state with precision,
but there is at least one series of thorough and
carefully conducted experiments reported by
Cotsonas (Note 3) in which external shock
(scrambled) as well as (negatively reinforcing) brain stimulation was used,
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
105
and where no preference emerged. A
conference paper by Douglass (Note 4) also
reports failure to find PSS in shuttle-box
situations, and perhaps most importantly,
following the Perkins et al. (1966) “factorsbehind-the-PSS” paper, the published
literature is substantially silent with respect to
positive and statistically assessable instances
of PSS with unmodifiable shock.3
Indeed, Arabian and Desiderato (1975)
reported what has, to date, been the most
convincing evidence for PUS, or for the
aversive properties of the signal. The relevant
group in this study chose (in a shuttle-box)
between a situation where each (scrambled)
shock was preceded by a 5-sec tone (i.e., signalled state) and one where the tone was
uncorrelated (i.e., unsignalled state). This
group developed over a 90% preference for
the latter condition (Arabian & Desiderato,
1975). However, the results from this procedure, taken as a whole, do not support an
overall PUS interpretation for at least two
reasons. First, the group choosing not to have
the 5-sec tone signal was still in a signalled
state with respect to 10-min light signal that
indicated whether or not (but not precisely
when) shocks would occur, and with respect
to this latter “nontemporal” (Furedy &
Biederman, 1976) signal another group
showed a clear PSS when given the choice
between having the nontemporal light signal
and not having it. Secondly, Collier (1977)
employing the same preparation though with
slightly different parameters (e.g., 10-rather
than 5-sec tone temporal signals) obtained
PSS outcomes even for the group given the
choice between the having and not having the
tone signal the precise time of shock onset. So
the Arabian-Desiderato (1975) PUS outcome
appears to be restricted to a rather
3
As recently as 1972, a published report uncritically
using unscrambled shock and finding moderate PSS
could still be found (French, Palestino, & Leeb).
narrow set of conditions, but this, of course,
is not to deny that the aversive properties of
the signal can be important in determining
whether there is PSS, PUS, or no preference.
There has, on the other hand, been one
equally strong apparent PSS effect reported
by Miller et al. (1974) with shocks delivered
through tail electrodes in order to rule out
modifiability factors. According to these
investigators the secret of their successful
“demonstration” (p. 273) of PSS was the great
number of training days given to their
subjects, in contrast to the relatively short
training sessions employed by such
investigators as Biederman and Furedy
(1973). However, Miller et al. (1974)
obtained their effect in only three animals
inasmuch as four others were eliminated for
removing the shock electrodes (Miller et al.,
1974, p. 274, n.l). It is therefore difficult to
evaluate the statistical reliability of the result.
Moreover, in a related investigation of the
amount-of-training factor with an adequate
(n= 14) sample of subjects, Biederman and
Furedy (1976a) trained rats for 50 days in the
shuttle-box with scrambled shock, and failed
to obtain any significant preference.4
Taken together, then, the results
4 Miller and associates have very recently published a
series of experiments wherein control of PSS appears to
have been gained in a symmetrical design (Miller,
Marlin, & Berk, 1977) in addition to having replicated
Biederman and Furedy*s (1976a) parameters, and
findings of no preference. The Miller et al. (1977)
evidence would seem to indicate that experimental
parameters strongly affect PSS or PUS outcomes.
While this new evidence is not consistent with the view
that PSS is a robust, readily obtainable phenomenon
(they report considerable intragroup variation), their
study may clearly become a major weight in the scale
balancing between PSS and PUS. The attempt to
balance PSS findings against PUS or no-preference
outcomes is, of course, hampered by the uncertain
number of no-preference or “negative** (i.e., PUS)
results which may remain unpublished because of the
general status of anti-paradigmatic results or results
which are merely neutral with respect to a prevailing
paradigm.
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
106
from those post-1966 studies of symmetrical
choice which did not permit unauthorized
modification through use of unscrambled
grids provided little if any evidential support
for the notion of PSS. During the same period,
two studies were reported (Biederman &
Furedy, 1973; Furedy & Biederman, 1976)
which directly examined the role of
scrambling in symmetrical choice experiments, and which showed both in terms of
preference behaviour itself, and in terms of
continuously
measured
amount
of
unauthorized
(unscrambled)
shock
modification, that: (a) PSS emerged only
when the shock was unscrambled, and (b)
there was a clear-cut connection between
modification and the PSS, although both
effects were weak, i.e., only partial
modification and PSS of the order of only
70%. The relative weakness of these
modification effects is important both because
it is consistent with Lockard’s (1963) failure
to find modification when it was measured
only on an all-or-none basis (cf. Furedy &
Biederman, 1976, Exp. HI), and because it
explains why previous investigators (with the
exception of Badia, Note 1) did not
immediately
remedy
this
confound.
Nevertheless, as shown in the BiedermanFuredy studies, the scrambling confound appears completely to account for the PSS effect
that can be obtained in the symmetrical-choice
(usually shuttle-box apparatus) situation.
Accordingly, these studies of the effect of
scrambling the grid provided additional
evidential problems for the PSS paradigm in
symmetrical choice designs.
From the paradigmatic view of PSS,
however, the greatest empirical shortcoming
of the symmetrical choice designs was that
they simply did not “work” very well as tools
to demonstrate PSS or the signal control of
behaviour. Some form of the symmetrical
choice design (e.g., the shuttle-box) is
logically ideal as a procedure to investigate
the relative strengths of the
”positive1”
(i.e.,
informational)
and
“negative” (i.e., conditioned-fear) properties
of signalling. However, if one is concerned to
demonstrate the PSS phenomenon, clearly
these procedures are inadequate, since, as we
have seen, even when unauthorized
modification is permitted, the preference
levels reached are typically of the order of
70%; moreover, the effect cannot be produced
in every individual subject but is found only
at the level of large sample averages. Even
Lockard’s (1963) more widely cited study (in
contrast to her less strong 1965 results) indicated that PSS assessed in this way showed
great “individual differences” which is one
feature of behaviour that is intolerable for the
demonstrational-paradigmatic approach.
The PSS, having attained full paradigmatic
status for most experimenters, and lacking a
truly powerful demonstrational method, was
dramatically bolstered by the asymmetricalchangeover procedure of Badia and his coworkers, a procedure which, despite its
methodological shortcomings (to be discussed
below), but because of its (seeming)
usefulness as a demonstrational tool, has
apparently displaced symmetrical-choice
preparations as the preferred measure of
preference.
THE ASYMMETRICAL-CHANGEOVER
PROCEDURE OF THE SEVENTIES: A
METHODOLOGICAL ANALYSIS
The method we will call asymmetrical
changeover procedure (ACP) as introduced
by Badia and co-workers (Badia, Coker, &
Harsh, 1973; Badia & Culbertson, 1972;
Badia, Culbertson, & Harsh, 1973) can be
said to have provided the new evidential basis
for current belief in animal PSS. Thus both
D’Amato (1974) and Seligman (1975) rely
heavily on this work to support the generality
and power of animal PSS, and the earlier
Lockard (1963) and Perkins et al. (1966)
papers are not mentioned. The reason for
this
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
107
preference for ACP on the part of reviewers
accepting the PSS paradigm is not hard to
find. As D’Amato states, in reference to one
of the later ACP papers (Badia, Culbertson, &
Harsh, 1973), “the magnitude of the
difference in aversiveness of the two types of
shock is indicated by the fact that in one study
rats chose signaled shock ... 2 to 3 times more
intense than unsignaled shock” (D’Amato,
1974, p. 93). The degree of “behavioural
control” manifested by the ACP seems indeed
impressive, especially when one recalls that
Lockard (1963), who seemed to be able to get
a 90% PSS (albeit with unscrambled shocks),
obtained a PUS when the signalled shock was
as little as 7% stronger than the unsignalled
shock (Lockard, 1965); (see our footnote 1).
Not only because of its apparent power, but
because of its relative complexity, the ACP is
worth diagramming in order to make clear the
procedures involved. In Fig. 1 we have
chosen to illustrate the first inescapable shock
experiment in the ACP series, i.e., Experiment II of Badia and Culbertson (1972).
As the bottom channel of Fig. 1 shows,
shocks were delivered on a variable-interval
response independent schedule.
The second channel shows the operant (CO),
which is the “changeover” (bar-press)
response, and, contingent on this response
(contingencies indicated by arrows) is a
“correlated stimulus” (top channel) which, in
that study, was a 3-min change in illumination
(light on for some animals, and off for others).
The arrangement is that if the changeover
response occurs (except for the 3-min period
following the last CO response), it is followed
immediately by the onset of the correlated
stimulus (as is illustrated in Fig. 1 by the first
and fourth responses). The effect of the
correlated stimulus is that while it is on, all
shocks are signalled, as shown in the
“warning signal” channel. Thus, in the
examples shown in Fig. 1, while the
correlated stimulus is on, following the first
changeover response, the first two responseindependent shocks which appear are
signalled (with a tone as the signal), but the
next two shocks are not signalled because the
correlated stimulus has terminated. Of course,
the absence of the correlated stimulus is ended
by the occurrence of the next changeover
response (#5 in Fig. 1) which, in turn, insures
that the next response-independent shock is
once again signalled. The “2-sec delay”
FIG. 1. Badia and Culbertson (1972, Exp. II) arrangements. For explanation see text. Note that following
onset of unsignalled shocks a 2-sec delay was in effect during which responses were ineffective in
producing the correlated stimulus and the warning signal.
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
108
refers to the arrangement where following
shock offset bar presses do not produce
correlated stimulus. This 2-sec delay is
intended to obviate the learning of spurious
bar press-shock termination contingencies.
At an intuitive level, this arrangement looks
relatively complicated compared with the
symmetrical choice shuttle-box (e.g., Lockard,
1963) procedures. However, given the strong
“signal control” over behaviour that has
impressed some reviewers, it appears that
there is an implicit attitude that the intuitive
complexities associated with ACP should not
be allowed to prevent its use as a dominant
index of rat PSS.5 We, however, do not agree
with this attitude, and have raised
methodological questions concerning ACP in
preliminary (Biederman & Furedy, 1973, p.
385) and conference (Furedy & Biederman,
Note 5) formats. More recently, in a technicalapparatus (instrumentation) and experimentaldesign critique of the ACP, we have detailed
six problems with the
5 We have elsewhere reported evidence to indicate
that the preparation does not meet scientific standards of
replicability (Biederman & Furedy, 1976b; 1977).
However, since this issue is experimentally soluble, we
shall not pursue it here, but shall, for the purpose of this
paper, assume that, contrary to our findings, operational
duplications of the ACP procedure would produce
behavioural replications of the changeover phenomenon
reported by Badia and Culbertson (1972). The present
problem has to do with difficulties in interpreting such
changeover phenomena (assuming them to be
replicable), difficulties which as indicated below arise
from methodological problems inherent in the procedure. On the other hand, the replicability problem is
of relevance for evaluating the relative empirical gravity
of the seven logical or methodological problems
presented in the text. A precise evaluation is possible
only with a replicable procedure where one can obtain
the basic (logically confounded) phenomenon and then
evaluate the empirical importance of each of the
confounds by removing each to see if that removal
affects the outcome. With an unreliable procedure like
the ACP such an empirical or experimental evaluation
of the empirical importance of the confounds is not
possible.
procedure (Biederman & Furedy, 1976c).
Accordingly, in what follows we shall, in the
interests of space, refer to this 1976 article in
all cases where our arguments are the same.
Our purpose here is to present a final version
of the methodological case against the ACP
procedure. In addition, we shall comment
briefly both on more recent empirical ACP
papers (e.g., Badia, Harsh, Coker, & Abbott,
1976) as well as on some methodological
arguments which have been offered (Badia &
Harsh, 1977a, 1977b) in support of the ACP
procedure as a measure of preference.
The methodological case against ACP as a
measure of preference is that there now seem
to be at least seven conceptual difficulties
associated with this procedure which
invalidate it as an index of PSS, or, for that
matter, as an index of preference between any
two stimulus classes (Biederman & Furedy,
1976c, p. 502). The earlier symmetricalchoice methods are free from all of these
difficulties, simply because an arrangement
like the shuttle-box (e.g., Lockard, 1963),
whatever other problems it may have, at least
provides the organism with a symmetrical
choice between the two states (signalled and
unsignalled) in a way that the ACP does not.
The first and most fundamental of ACP’s
problems is that of choice response
asymmetry. To summarize what we have
detailed elsewhere (Biederman & Furedy,
1976c), the problem is that radically different
responses (bar-press and withholding-of-barpress) lead to the two states (signalled and
unsignalled, respectively) for which unbiased
preference is intended to be measured (see
also Fig. 1). An empirical outcome which
conveniently illustrates this problem is one
where, in the arrangements depicted in Fig. 1,
most subjects show near zero percent
changeover. If the ACP were a valid index of
preference, such a result
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
109
would be interpretable as very strong
preference for unsignalled shock (PUS).
However, as we noted in connection with
data which did yield such low changeover
results, the “more appropriate interpretation
of these data is, of course, that the CO
response simply had a low operant rate and
that asymmetrical changeover is not a valid
measure of choice” (Biederman & Furedy,
1976b, p. 422).
A second and related problem in ACP
arises from an asymmetry of responsestimulus contingencies between the behaviour
(bar press) producing the signalled state and
the behaviours (summarized by the term “nobar-press”, but consisting of an indefinite
number of heterogeneous behaviours such as
scratching,
grooming,
freezing,
etc.)
producing the unsignalled state. Although we
have described a brief version of this problem
previously (Biederman & Furedy, 1976c, p.
508), that account requires clarification in
some respects. Specifically, the problem is
that the behaviour taken to index signalled
preference (in Fig. 1 see changeover
responses #1 and #5) is sometimes
immediately followed by a clear consequence
— the change from unsignalled to signalled
state — so that in this case there is a relatively
clear contingency between a homogeneous
response (bar-press) and a consequent
stimulus (see correlated stimulus channel in
Fig. 1). On the other hand, the no-bar-pressfor-three-minutes behaviours taken to index
unsignalled preference do not have a parallel
contingency between a single type of
response and an immediately consequent
stimulus. Thus, in Fig. 1, the behaviour
occurring at the end of the no-bar-press-forthree-minutes criterion that caused the first
change back to the unsignalled state
(correlated stimulus off) could be quite
different from that which immediately
preceded the second change back to the
unsignalled state. A response-stimulus
contingency can be specified in Fig. 1 for the
purported index of signalled preference (see
top two channels, changeover responses #1
and #5) but no similar specification can
possibly be made for any index of unsignalled
preference in ACP.
The third problem, termed indicatorstimulus ambiguity, again arises from the
asymmetrical nature of ACP, but our previous
account (Biederman & Furedy, 1976c, pp.
508-510) of both the logical nature of and
empirical evidence for this confound is
sufficiently detailed as to require only a very
brief summary here. The nature of the confound can be gleaned from the correlated
stimulus channel in Fig. 1. This 3-min
consequence of the changeover response
(second channel in Fig. 1) is intended by the
experimenter to serve as an indictor that
shocks will be signalled by the “warning
signal” (third channel in Fig. 1). However, the
correlated stimulus may produce bar pressing
simply because of its properties qua stimulus,
i.e., the stimulus change itself may be
reinforcing. Empirical evidence for the
confound is available by analysis of the
results themselves of the Badia and
Culbertson (1972, cf. their Fig. 3, p. 469, and
Biederman & Furedy, 1976c, pp. 508-509), as
well as a study by Biederman and Furedy
(1973; cf. Biederman & Furedy, 1976c, pp.
509-510). The evidence from the former
source is particularly worth noting because in
all subsequent studies, except for one by
Lewis and Gardner (1977), Badia and his
associates have omitted the control groups
(correlated stimulus alone with no signal
consequent upon lever press, which produced
a higher level of changeover performance; no
correlated stimulus but signals still
consequent upon lever press, which produced
little or no changeover performance, cf. Badia
and Culbertson, 1972, Fig. 3) that make it
possible to obtain this type of evidence.
However, of course, the methodological
problem
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
110
that is due to indicator-stimulus ambiguity
exists independently of the amount of
illustrative empirical evidence that may be
available.
The fourth and fifth problems, labelled
respectively dependent-variable measure and
asymmetry of interpretation have been
sufficiently described (Biederman & Furedy,
1976c, pp. 510-511). The former refers to the
fact that the dependent variable — percent
time spent on signalled side — is confounded
by the operant rate of bar pressing in a way
that is not the case in any symmetrical-choice
preparations (e.g., Lockard, 1963). The latter
states the difficulty that the ACP design,
unlike a symmetrical-choice preparation
(SCP) is less unequivocal when un-signalled
or no preference occurs than when signalled
preference occurs.
The sixth problem with ACP as a measure
of preference arises from the very long-term
changeover or “preference” training that
animals are subjected to, and may be termed
for reasons to be given, the filet mignon to
hamburger
or
satiation
confound.
Specifically, unlike the symmetrical-choice
designs where, in periods of no preference the
animal is exposed to an equal extent to the
signalled and unsignalled states, the ACP is so
designed that at low rates of changeover or
“preference”, the animal is exposed for a
longer time to the unsignalled than the
signalled state. One relevant analogy is a test
of preference in humans between (wellprepared) filet mignon and cheap, ill-prepared
hamburger, designed as follows: the filet is
provided at regular response-independent
intervals (i.e., analogous to no-signal),
whereas the hamburger (analogous to signal)
is contingent on some operant. After 50 days
of filet mignon, the operant response producing (“changing over for”) hamburger might
well appear, because of selective satiation
with steak. It is even possible that humans
might “prefer” a ham-
burger diet for the next few days at levels
close to 90%. To conclude on such grounds
that hamburger was preferred to filet mignon
would be clearly unwarranted, and would be
acceptable only to hamburger enthusiasts who
wanted to demonstrate the preference-forhamburger
phenomenon,
rather
than
investigate whether steak or hamburger was
preferred. A valid investigation of that meaty
problem, of course, would require a
symmetrical-choice preparation rather than
some analogue of the animal ACP.
The seventh and last problem with ACP
relates to the later and more spectacular
reports (Badia, Coker, & Harsh, 1973; Badia,
Culbertson, & Harsh, 1973) in which the ACP
apparently showed that rats prefer much
stronger, denser, and longer shocks if these
were signalled. The methodological problem
with these ACP studies is one of response
fixation, and we have previously both
described and cited evidence for the existence
of this difficulty with the ACP design
(Biederman & Furedy, 1976c, pp. 511-512).
In that paper we also suggested that although
the various problems are linked to the
asymmetricality of the procedure, they are
also independent in the sense that all seven
must be rejected for the ACP to be shown to
be a valid index of preference. The present
seven problems that we raise probably vary
somewhat in their perceived gravity but we
think that rejection of all assertions is unlikely
even for those for whom PSS has, indeed,
become a paradigm.6
6 Nevertheless, for those workers who would want to
reject all seven of the above methodological difficulties
as inconsequential, and who would also not wish to
accept the replicability problems with the ACP
procedure (footnote 5), there is also a set of technical
problems which need to be considered with ACP as it
has been used in the Bowling Green Laboratory. These
problems, which are essentially apparatus difficulties
associated with the modified Foringer chamber used in
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
111
Forms of the ACP have continued to be
reported by Badia and his associates (Badia &
Harsh, 1977a, 1977b; Badia, Harsh, & Coker,
1975; Badia et al., 1976; Culbertson & Badia,
1977; Harsh & Badia, 1974, 1975, 1976;
Lewis & Gardner, 1977). These later studies,
indeed, seem generally to be less enthusiastic
about the strength of PSS. For example,
Harsh and Badia (1976) report that what they
take to be PSS occurs with a 150-sec period
during which shocks are signalled but not
with a 45-sec period. A similar class of
finding with respect to shock intensity had
already been reported with low shock
intensity not yielding PSS (Harsh & Badia,
1975). Also, and perhaps more importantly,
there has been no repetition of the apparently
spectacularly strong PSS reports of
changeover for much longer, stronger, and
denser shocks (Badia, Coker, & Harsh, 1973;
Badia, Culbertson, & Harsh, 1973). All this
might be seen by some as indicating that the
parameters of PSS are being established by
these later ACP experiments. However, the
methodological difficulties we have raised in
connection with the original ACP studies
apply with equal force to the later ACP examples because the difficulties are in the form of
logical objections.
As we have previously pointed out, it
seems obvious that designs that are intended
to measure preference validly should do so
(Biederman & Furedy, 1976c), and there are
now at least seven methodological reasons
why the ACP does not. We take the fact that
ACP continues to be used and reported to
illustrate only the strength of the PSS as a
paradigm, rather than any validation of ACP
as an adequate test of
footnote 6 continued
the Bowling Green Laboratory, raise the question of
modifiability, and suggest that the shocks used in the
ACP experiment which were characterized as
“inescapable” were, in fact, not so (Biederman &
Furedy, 1976c, pp. 502-507; cf. especially photograph
in Fig. 3 and recordings in Fig. 4).
preference. That strength of belief in the PSS
paradigm has been nicely, if perhaps
unintentionally, illustrated recently by Lewis
and Gardner (1977) who report a replication
of the Badia and Culbertson (1972) ACP
results, and, consistent with our footnote 5 in
this paper, we shall for now assume that
replicability of this particular ACP experiment
can, despite our contrary findings (Biederman
& Furedy, 1976b), be achieved. However,
Lewis and Gardner also used a symmetrical
changeover condition (where the same lever
press produced both the signalled and the
unsignalled shock on an alternating basis),
and found PSS only in the two animals who
were
previously
exposed
to
the
(methodologically
confounded)
ACP
arrangement. The verdict of these authors on
the results from three animals only subjected
to
an
(un-confounded)
symmetrical
changeover is that because those three rats
“failed to develop a preference for signaled
shock ... this particular symmetrical procedure
appears inferior to the ACP changeover
procedure” (Lewis & Gardner, 1977, p. 138,
emphasis ours). When results inconsistent
with the notion of PSS are categorized as
“failures”, and the relative merits of designs
are apparently judged in terms of how well
they “succeed” in producing results that can
be interpreted as PSS, it is clear that a paradox
has indeed been transformed into a paradigm.
Most recently Badia and Harsh (1977a,
1977b) have offered some arguments against
our earlier suggestions that there are
difficulties with ACP as a measure of PSS or,
for that matter, of anything else. In this
connection it is important to focus discussion
on the PSS phenomenon itself as it was
intended to be investigated in the original
inescapable shock study of Badia and
Culbertson (1972, Exp. II), rather than cloud
what is already a sufficiently complex and
controversial issue by analysing in detail
numerous other papers
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
112
listed by Badia and Harsh (1977a, p. 14). Of
these cited papers some do not contrast only
signalled with unsignalled shock, but involve
contrasting
fixedand
variable-time
schedules (e.g., Badia et al., 1975), others do
not involve ACP at all, but a symmetricalchoice procedure (e.g., Safarjan & D’Amato,
Note 6), and yet others varied such
parameters as the intershock interval and
reported differential changeover as a function
of that variation (e.g., Badia et al., 1976). We
agree that probably “none of the factors
described by Furedy and Biederman can
account for these data” (Badia & Harsh,
1977a, p. 14), but that seems hardly the point.
As we have indicated above, the
methodological validity of the ACP as a
measure of preference has to be rejected if
only one of the seven arguments against it is
granted. The arguments do not depend on
their validity for their ability to fully account
for even the original Badia and Culbertson
(1972, Exp. II) study which was analysed;
their validity certainly is not contingent on
being able to be consistent with or account
for the various other later claims made on the
basis of later ACP papers, because, if at least
one of our seven arguments are granted, these
papers, like the one originally analysed, do
not measure PSS.
CONCLUSIONS
We have tried to show that an analysis of
the data base and
The fate of Mowrer’s (1947) version of two-factor
theory is another story (see Hernstein, 1969) and the
current unpopularity of WS-termination as an
explanation of avoidance learning clearly strengthens
the present readiness to uncritically accept PSS as
paradigmatic. The avoidance literature, which is quite
distinct from the PSS evidence, has had a solid
empirical basis relatively free from the gravitational
influences of major theoretical masses. For example,
Bolles, Stokes, and Younger (1966) found that while
WS termination has a role in avoidance, it is not the
critical variable; D’Amato, Fazzaro, and Etkin (1968)
suggested that WS “termination (on avoidance trials)
facilitates avoidance conditioning not because it results
in fear reduction but
7
methodological status of PSS from 1959 to
1976 fails to support the notion that PSS is a
robust phenomenon in rats.7 Moreover, it
seems even more clear that any paradigmatic
treatment of PSS, as if the converse
phenomenon (PUS) was in some sense
paradoxical, is an inadvisable way to study
the effects of signalling unmodifiable noxious
events. In particular, it appears that the ACP
which has been used recently to support the
paradigmatic status of rat PSS is itself a
paradigm that is open to methodological (as
well as replicational and technical) problems
which render ACP-based results questionable
for determining the preference issue. That
issue, more generally, needs to be considered
in a sceptical manner rather than in a
“demonstrational” (Furedy, 1975) manner
wherein not only the positive but also the
negative aspects of signalling are recognized.
As an experimental procedure signalling
shock is of considerable interest because it
allows the study, under relatively precisely
controlled laboratory conditions, of how
organisms cope with noxious unmodifiable
events. It is important to recall that the value
of this procedure is not dependent on being
able to “achieve” robust PSS results. Even if
PSS turns out to be an unstable effect
(because of the aversive properties of the
signal), it will still be of interest to study the
functions of the various
because it serves as an excellent discriminative cue for
US avoidance** (p. 41). Thus, termination of WS might
enhance avoidance by providing feedback (information)
to the rat that US will not occur. Bolles and Grossen
(1969) suggest that the WS in avoidance may function
in a way similar to the SD of appetitive discrimination
problems (cf. O’Neill & Biederman, 1974). Thus,
informational properties now attributed to the WS in
avoidance experiments are readily transferred to the inescapable, unavoidable PSS situation, despite the great
functional (and psychological) differences in roles of
the WS in avoidance learning and of the WS associated
with inescapable, unmodifiable, unavoidable shock.
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
113
physiological anticipatory responses elicited
by the signal from the point of view of
determining which of these lessen and which,
perhaps, heighten the impact of the signalled
noxious event. One hopes, of course, that
these “under-the-skin” events will show patterns that can be related to the behavioural
results represented by preference outcomes.
What is clear, however, is that neither type of
phenomena should be approached from a
paradigmatic-demonstrational point of view.
REFERENCE NOTES
1. BADIA, P. Intermittent immediate punishment
preceded by a warning signal as determiners of
choice. Un published MA thesis, Kent State
University, 1959.
2. PERKINS, C. C, JR. Personal communication,
1969.
3. COTSONAS, P. M. Preference for signaled
over
unsignaled
aversive
stimulation.
Unpublished MA thesis, University of North
Carolina, 1972.
4. DOUGLASS, W. K. Some determinants of the
preference for signaled shock. Paper presented
to Eastern Psychological Association meetings,
April 1969.
5. FUREDY, J. J., & BIEDERMAN, G. B.
Methodological problems in evaluating rat
preference for signaled or un-signaled shock.
Paper presented to meeting of Psychonomic
Society, November 1974.
6. SAFARJAN, W. R., & D’AMATO, M. R.
Discriminated shock-free periods as the
controlling variable in preference for signaled
shock.
Paper
presented
to
Eastern
Psychological Association meeting, New York
1975.
REFERENCES
ARABIAN, J. M., & DESIDERATO, O. Preference for
signaled shock: A test of two hypotheses.
Animal Learning and Behavior, 1975, 3, 191195.
AVERILL, J. R., & ROSENN, M. Vigilant and
nonvigilant
coping
strategies
and
psychophysiological stress reactions during the
anticipation of electric shock. Journal of
Personality and Social Psychology, 1972, 23,
128-141.
BADIA, P., COKER, C. C, & HARSH, J. Choice of
higher density signaled shock over lower
density unsignaled shock. Journal of the
Experimental Analysis of Behavior, 1973, 20,
47-55.
BADIA, P., & CULBERTSON, S. The relative
aversiveness of signaled vs. unsignaled
escapable and inescapable shock. Journal of
the Experimental Analysis of Behavior, 1972,
17, 463-471.
BADIA, P., CULBERTSON, S., & HARSH, J. Choice
of longer or stronger signaled shock over
shorter or weaker unsignaled shock. Journal of
the Experimental Analysis of Behavior, 1973,
79, 25-32.
BADIA, P. & HARSH, J. Preference for signaled
over unsignaled shock schedules: A reply to
Furedy and Biederman. Bulletin of the Psychonomic Society, 1977, 10, 13-16. (a)
BADIA, P., & HARSH, J. Further comments
concerning preference for signaled shock
conditions. Bulletin of the Psychonomic
Society, 1977, 10, 17-20. (b)
BADIA, P., HARSH, J., & COKER, C. C. Choosing
between fixed time and variable time shock.
Learning and Motivation, 1975, 6, 264-278.
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
114
BADIA, P., HARSH, J., COKER, C. C, & A3BOTT, B.
Choice and the dependability of stimuli that
predict shock and safety. Journal of the
Experimental Analysis of Behavior, 1976, 26,
95-111.
BERLYNE, D. E. Conflict, arousal, and curiosity.
New York: McGraw-Hill, 1960.
BIEDERMAN, G. B., & FUREDY, J. J. The
preference-for-signalled-shock phenomenon:
Signalling shock is reinforcing only if shock is
modifiable. Quarterly Journal of Experimental
Psychology, 1970, 22, 681-685.
BIEDERMAN, G. B., & FUREDY, J. J. Preference-forsignalled-shock phenomenon: Effects of shock
modiliability and light reinforcement. Journal
of Experimental Psychology, 1973, 100, 380386.
BIEDERMAN, G. B., & FUREDY, J. J. The
preference-for-signalled-shock phenomenon:
Fifty days with scrambled shock in the shuttlebox. Bulletin of the Psychonomic Society,
1976, 7, 129-132. (a)
BIEDERMAN, G. B., & FUREDY, J. J. Operational
duplication without behavioral replication of
changeover for signaled inescapable shock.
Bulletin of the Psychonomic Society, 1976, 7,
421-424. (b)
BIEDERMAN, G. B., & FUREDY, J. J. Preference for
signaled shock in rats? Instrumentation and
methodological errors in the archival literature.
Psychological Record, 1976, 26, 501-514. (c)
BIEDERMAN, G. B., & FUREDY, J. J. The
preference-for-signalled-shock phenomenon:
Reliability and sensitivity of asymmetrical and
symmetrical changeover procedures in the
Skinner box. Australian Journal of Psychology,
1977, 29, 111-124.
BOLLES, R. C, & GROSSEN, N. E. Effects of an
informational stimulus on the acquisition of
avoidance behavior in
rats. Journal of Comparative and Physiological
Psychology, 1969, 68, 90-99.
BOLLES, R. C, STOKES, L. W., & YOUNGER, M. S.
Does CS termination reinforce avoidance
behavior? Journal of Comparative and
Physiological Psychology, 1966, 62, 201-207.
BRADY, J. V., & HUNT, H. F. An experimental
approach to the analysis of emotional behavior.
Journal of Psychology, 1955, 54, 345-352.
BROWN, J. W., & JACOBS, A. Role of fear in
motivation and acquisition of responses.
Journal of Experimental Psychology, 1949, 34,
747-749.
BRUSH, F. R., BRUSH, E. S., & SOLOMON, R. L.
Traumatic avoidance learning: The effect of
CS-UCS interval with a delayed conditioning
procedure. Journal of Comparative and
Physiological Psychology, 1955, 48, 285-293.
CANTOR, M. B., & LOLORDO, V. M. Reward value
of brain stimulation is inversely related to
uncertainty about its onset Journal of
Comparative and Physiological Psychology,
1972, 79, 259-270.
CHURCH, R. M, & SOLOMON, R. L. Traumatic
avoidance learning: The effects of delay of
shock termination. Psychological Reports,
1956, 2, 357-368.
COLLIER, A. C. Preference for shock signals as a
function of the temporal accuracy of the
signals. Learning and Motivation, 1977, 8,
159-170.
CULBERTSON, S., & BADIA, P. Changeover from
unsignaled to signaled avoidance as a function
of the changeover period duration. Bulletin of
the Psychonomic Society, 1977, 9, 159-162.
D’AMATO, M. R. Derived motives. Annual Review
of Psychology, 1974,25, 83-106.
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
115
D’AMATO, M. R., FAZZARO, J., & ETKIN, M.
Anticipatory responding and avoidance
discrimination as factors in avoidance
conditioning. Journal of Experimental
Psychology, 1968, 77, 41-47.
DESIDERATO, O., & NEWMAN, A. Conditioned
suppression produced in rats by tones paired
with escapable or inescapable shock. Journal
of Comparative and Physiological Psychology
1971, 77,427-435.
DINSMOOR, J. A. Punishment: I. The avoidance
hypothesis. Psychological Review, 1954, 61,
34-46.
ESTES, W. K., & SKINNER, B. F. Some
quantitative properties of anxiety. Journal of
Experimental Psychology, 1941, 29, 390-400.
FRENCH, D., PALESTINO, D., & LEEB, C.
Preference for warning in an unavoidable
shock situation: Replication and extension.
Psychological Reports, 1972, 30, 12-1 A.
FUREDY, J. J. An integrative progress report on
informational control in humans: Some
laboratory findings and methodological claims.
Australian Journal of Psychology, 1975, 27,
61-83.
FUREDY, J. J., & BIEDERMAN, G. B. Preferencefor-signaled-shock phenomenon: Direct and
indirect evidence for modifiability factors in
the shuttle-bos. Animal Learning and Behavior,
1976, 4, 1-5.
FUREDY, J. J., & DOOB, A. N. Autonomic
responses and verbal reports in further tests of
the
preparatory-adaptive-response
interpretation of reinforcement. Journal of
Experimental Psychology, 1971, 89, 258-264.
FUREDY, J. J., & DOOB, A. N. Signaling
unmodifiable shocks: Limits on human
informational cognitive control. Journal of
Personality and Social Psychology, 1972, 21,
111-115.
HARSH, J., & BADIA P. A concurrent assessment
of the positive and negative properties of a
signaled shock schedule. Animal Learning and
Behavior, 1974, 2, 168-172.
HARSH, J., & BADIA, P. Choice of signalled and
unsignalled shock as a function of shock
intensity. Journal of the Experimental Analysis
of Behavior, 1975, 23, 349-355.
HARSH, J., & BADIA, P. A temporal parameter
influencing choice between signalled and
unsignalled shock schedules. Journal of the
Experimental Analysis of Behavior, 1976, 25,
327-333.
HERRNSTEIN, R. J. Method and theory in the study
of avoidance. Psychological Review, 1969, 76,
49-69.
HYMOWITZ, N. Effects on responding of mixed
and multiple schedules of signaled and
unsignaled response-dependent electric-shock
deliveries. Journal of the Experimental
Analysis of Behavior, 1976, 25, 321-326.
KAMIN, L. J. The effects of termination of the CS
and avoidance of the US in avoidance learning.
Journal of Comparative and Physiological
Psychology, 1956, 49, 420-424.
KAMIN, L. J. The gradient of delay of secondary
reward in avoidance learning. Journal of
Comparative and Physiological Psychology,
1957, 50, 445449. (a)
KAMIN, L. J. The gradient of delay of secondary
reward in avoidance learning tested on
avoidance trials only. Journal of Comparative
and Physiological Psychology, 1957, 50, 450456. (b)
KNAPP, R. K., KAUSE, R. H., & PERKINS, C. C., JR.
Immediate vs. delayed shock in T-maze performance. Journal of Experimental Psychology,
1959, 55, 357-362.
LEWIS, P., & GARDNER, E. T. The reliability of
preference for signaled
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
116
shock. Bulletin of the Psychonomic Society,
1977, 9, 135-138.
LOCKARD, J. S. Choice of a warning signal or no
warning signal in an unavoidable shock
situation. Journal of Comparative and
Physiological Psychology, 1963, 3, 526-530.
LOCKARD, J. S. Choice of a warning signal or
none in several unavoidable-shock situations.
Psychonomic Science, 1965, 3, 5-6.
MACDONALD L., & BARON, A. A rate measure of
the relative aversiveness of signaled vs.
unsignaled shock. Journal of Experimental
Analysis of Behavior, 1973, 19, 33-38.
MILLER, N. E. An experimental investigation of
acquired drives. Psychological Bulletin, 1941,
38, 534-535.
MILLER, N. E. Studies in fear as an acquirable
drive: I. Fear as motivation and fear-reduction
as reinforcement in the learning of a new
response. Journal of Experimental Psychology,
1948, 38, 89-101.
MILLER, R. R., DANIEL, D., & BERK, A. M.
Successive reversals of a discriminated
preference for signaled tailshock. Animal
Learning and Behavior, 1974, 2, 271-274.
MILLER, R. R., MARLIN, N. A., & BERK, A. M.
Reliability and sources of control of
preferences for signaled shock. Animal
Learning and Behavior, 1977, 5, 303-308.
MOWRER, O. H. On the dual nature of learning: A
reinterpretation
of
“conditioning”
and
“problem solving”. Harvard Educational
Review, 1947, 17, 102-148.
MOWRER, O. H., & AIKEN, E. G. Continuity vs.
drive-reduction in conditioned fear: Temporal
variation in conditional and unconditional
stimulus. American Journal of Psychology,
1954, 67, 26-38.
MOWRER, O. R, & SOLOMON, L. N.
Contiguity versus drive reduction in
conditioned fear: The proximity and abruptness
of drive-reduction. American Journal of
Psychology, 1954, 67, 15-25.
O’NEILL, W., & BIEDERMAN, G. B. Avoidance
conditioning as a function of appetitive
stimulus pretraining: Response and stimulus
transfer effects. Learning and Motivation,
1974, 5, 195-208.
PERKINS, C. C, JR. An analysis of the concept of
reinforcement. Psychological Review, 1968,
75, 155-172.
PERKINS, C. C, JR., LEVIS, D. J., & SEYMANN, R.
Preference for signal-shock vs. shock-signal.
Psychological Reports, 1963, 13, 735-738.
PERKINS, C. C, JR., SEYMANN, R. G., LEVIS, D. J.,
& SPENCER, H. R., JR. Factors affecting
preference for signal-shock over shock-signal.
Journal of Experimental Psychology, 1966, 72,
190-196.
SCHOENFELD, W. N. An experimental approach to
anxiety, escape, and avoidance behavior. In P.
H. Hoch & J. Zubin (Eds.), Anxiety. New
York: Grune and Stratton, 1950.
SELIGMAN, M. E. P. Helplessness. San Francisco:
Freeman, 1975.
SIDMAN, M. Avoidance conditioning with brief
shock and no exteroceptive warning signal.
Science, 1953, 118, 157-158.
SOLOMON, R. L., & BRUSH, E. S. Experimentally
derived conceptions of anxiety and aversion. In
M. R. Jones (Ed.), Nebraska Symposium on
Motivation. Lincoln: University of Nebraska
Press, 1956.
SUBOSKI, M. D., BRACE, T. G., JARROLD, L. A.,
TELLER, K. J., & DIETER, R. Interstimulus
interval and time estimation in ratings of
signaled shock aversiveness. Journal of
Experimental Psychology, 1972, 2, 407-415.
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
117
VERHAVE, T. Technique for the differential
reinforcement of rate of avoidance responding.
Science, 1959, 129, 959-960.
WEISS, J. M. Somatic effects of predictable and
unpredictable shock. Psychosomatic Medicine,
1970, 32, 397408.
WEISS, J. M. Effects of coping behavior
in different warning signal conditions on stress
pathology in rats. Journal of Comparative and
Physiological Psychology, 1971, 77, 1-13. (a)
WEISS, J. M. Effects of punishing the coping
response (conflict) on stress pathology in rats.
Journal of Comparative and Physiological
Psychology 1971, 77, 14-21. (b)
Received 7 June 1978.
Australian Journal of Psychology Vol 31, No. 2. 1979, pp.101-118
Download