Irrelevant stimuli present all or half of the time as subsequent discriminanda
Irrelevant stimuli present all or half of the time
as subsequent discriminanda
1
SALLY E. SPERLING
UNIVERSITY OF CALIFORNIA, RIVERSIDE
Irrelevant stimuli each associated with a 50% reinforcement schedule were either present on all or on half of initial
discrimination training trials. For different groups in test
problem training a previous ly irrelevant stimulus was either
the positive or the negative discriminandum, with a novel
stimulus as the other discriminandum, or both discriminanda
were novel stimuli. No effect due to novelty was demonstrable. Ss were retarded in test learning if their test negative
discriminandum had been present on half of the i;1itial acqUisition trials; no other groups differed from controls.
It is now clear that stimuli may acquire discriminative control over responding even though the reinforcement schedules associated with their presence differ
from the conventional 100% or 0% (e.g., Babb, 1956;
Jeeves & North, 1956; Sperling, 1962, for previously
irrelevant cues as subsequent discriminanda; Erlebacher, 1963; Grosslight, Hall, & Scott, 1954; Kendler
& Lachman, 1958; North & McDonald, 1962; Pavlik &
Born, 1963, for discrimination learning with intermittent
schedules of reinforcement for responses to the positive
discriminandum, the negative discriminandum, or both).
However, the specific conditions under which irrelevant
stimuli become discriminative are not specifiable from
these data; the present experiment was designed to
examine some of these conditions.
Method
The Ss were 46 naive male hooded rats, approximately 90 days of age. The data for two additional Ss
were discarded since they died during the experiment.
The apparatus was two elevated platforms enclosed
in a cubicle compartment divided by a partition into
starting and stimulus sections. When S broke an
infrared light beam at the terminal end of the starting
platform, a 100-w bulb over the stimulus compartment
was turned on and a timer was started which stopped
when S stepped off the stimulus platform.
A 20-v light bulb centered on a swinging door at
the terminal end of the platform could be either on
(L) or off (D). Four different stimulus platforms
covered with 1 in. Mystic Tape were used: Checkerboard (C) made of alternating black and white squares;
Stripe (S) made of alternating longitudinally laid black
and white tape; Zigzag (Z) made of alternating black
and white diagonals laid from the longitudinal center;
and Gray (G) made of longitudinally laid gray tape.
The ratio of black to white in the areas of the first
three platforms was about equal within and among
platforms.
A 23 hr. deprivation schedule was maintained through-
Psychon. Sci., 1966, Vol. "
out 12 days of pretraining and the experiment. All Ss
received initial training on a (L + D-) discrimination
problem to a criterion of not more than one trial
in a day on which the latency of response in the
presence of the negative discriminandum was shorter
than any latency of response in the presence of the
positive discriminandum. There were six positive and
six negative trials per day in random order on a 10-min.
intertrial interval. A food cup was baited with two 94
mg Noyes pellets on positive trials; it was not present
on negative trials.
Subgroups of Ss had different stimulus platforms
present during this training. For Group C (N = 16) the
C platform was present on each trial; for Group Z
(N = 14) the Z pattern was present on each trial; for
Group C Z (N = 8) the C platform was present on half of
the L and half of the D trials each day (three each)
and the Z platform was present on the other trials;
for Group CS (N = 8) the arrangement was the same
except that the S platform was used in place of the Z.
On the day following their criterion day on the L
vs. D problem, each S began test problem learning.
The swinging door containing the light bulb was removed
from the apparatus. All procedures remained the same,
and training was given to the acquisition criterion.
Half of the Ss were presented with a problem in which
C was the positive discriminandum and G was the negative (C + G-); for the other half of the Ss S was positive
and C negative (S+C-). These groups were constituted
as follows: Half of training Group C went into each test
problem (Cic+G-) and (C/s+c-); half of training
Group Z went into each problem; (Z/c+G-) and
(Z/s+c-); training Group CZ was given the S vs.
C problem (CZ/s+c-); and training Group CSwas given
the C vs. G problem (CS/s+G-).
Results
Mean trials to criterion for the two problems are
presented in Table 1, ordered with respect to test
l~arning. There were no differences among the six
experimental groups on trials to criterion on the L vs.
D problem (F=2.14; df=5/40; p> .05). An analysis of
Table 1. Mean Trials to Criterion on Acquisition and Test Problems
Groups
Proble"'(CZlS+C_)(C/C+G_)(CS/C+G_)(Z/S+C_)(C/S+C_)(Z/C+G-)
Acquisition
of L+D- 106.5
Test
84.0*
*
132.0
54.0
115.0
52.5
82.3
51.4
94.5
48.0
108.0
39.4
This mean is significantly different from each of the others in the
row (p <.OJ)
195
variance for test problem learning yielded F= 9.10;
df=5/40; p< .001. The differences between these means
were tested by 99% confidence intervals using Tukey's
test. Group (CZ/S + C-) was significantly retarded in
test problem learning compared to all of the other
groups which did not differ from each other.
Analyses of transformed latency scores (log lat)
indicated that terminal acquisition response levels in
the presence of the C and Z stimuli did not differ
among the group. There was also no apparent difference
in latency of response to c on the first day of test
problem training (both Fs were less than one), even
though C was a novel stimulus for two of these groups(Z/c+c-) and (Z/s+c-). Mean log lat for Sand G
on the first test day differed neither from each other
(F = 1.27; df = 5/40; p> .20) nor from mean log lat for C
(difference t= 2.54; df= 5; p> .05).
Mean log lat within each group on the second test
day indicated that only Group (CZ/S + C-) was still
responding faster on negative than on positive trials.
Discussion
If certain conditions of prior experience with a stimulus lead to the acquisition of discriminative control
over responding by that stimulus, differences should
be observable in the performance of groups who have
had the necessary prior experience compared to those
who have not. Neither Group (Z/C + G-) nor Group
(Z/s+ C-) had had any prior exposure to the two sets
of test problem discriminanda. Group (CZ/S + C-) was
the only experimental group which differed from them
in test learning, and it was significantly slower.
Resistance to extinction in the presence of C seems
to account for the retardation of this group which
persisted longer in responding faster to C than any
other group. The discriminative cor,trol over responding
exhibited in resistance to extinction must have been
acquired during initial training as a consequence of
C being an intermittent feature of (...truncated)
