The learning of bidirectional conditional relations (symmetry) has rarely been demonstrated in
non-human. Recently, three studies presented positive data from behavioral repertoires of
symmetry in pigeons. These studies indicated some variables as possible determinants of the
emergence of symmetry: 1) Intermixed training of arbitrary and identity relations at the same
time and with the same set of stimuli, 2) Non-exposition to prior training of any kind of
relationship with same stimuli, and 3) The use of the successive conditional discrimination
(Go/No-go). This study aimed to determine the need for the use of Go/No-go, using the other
variables (intermixed training and no prior exposure), to obtain the emergence of symmetry.
This study evaluated in two capuchin monkeys (Cebus apella) the learning of arbitrary and
identity relations of stimuli presented in the same experimental session and the emergence of
symmetry, using two procedures. One subject (M18) was trained in Go/No-go task with
mixed arbitrary and identity trials, with the same set of stimuli. The subject went through two
Go/No-go training that have not been successful in establishing conditional control. Another
subject (M27) was trained in a 0-delayed matching-to-sample task with mixed arbitrary and
identity trials, with the same set of stimuli too. This second procedure aimed to determine the
need to use the procedure Go/No-go to obtain the emergence of symmetry. Training sessions
were composed of eight identity trials (four relations) and eight arbitrary trials (two relations),
presented in random sequence. All trials had three choices as comparisons. Subsequently,
M27 was submitted to three tests (all of the trials had programmed reinforcement). The
intermixed baseline training (identity and arbitrary) occurred in 14 sessions. The first test of
symmetry was a session of eight trials for each of the four relations of identity, two arbitrary,
and two test trials in a total of 64 trials, and the performance criterion was seven correct trials
out of eight for each relation. M27 had 2 errors in a identity relation and reached the criterion
for all the others, including those of symmetry. To determine whether there was a consistency
of stimulus control, the second test was made in which the choices between the comparisons
could no longer be conditional to the model, since was used one single new stimulus as
sample at all time in both baseline and test trials. M27 made three errors global. This result
may indicate that some kind of unprogrammed stimulus control, developed: as the same two
pairs of stimuli alternated in the S- function in all trials, M27 may have learned to reject the
two pairs of stimuli independently of the sample presented. In order to verify if indeed there
was no control for selection according to the sample, it was ran a test with only two
comparisons with the same settings of the first test. M27 made six errors in 32 identity
baseline trials, six in 16 arbitrary trials and six in the 16 symmetry trials. The correct
responses were therefore above chance level, which partly confirms the hypothesis above and
may indicate the presence of a mixed stimulus control between conditional and discriminative
stimuli for comparison.
|
