- The purpose of this study was to test Siegel's utility model of
choice behavior under conditions of verbal reinforcement, in a nursery
school population of both sexes. Two major hypotheses were
tested: The ordinal hypothesis stated that under conditions of increased
value of the payoff subjects' stable-state strategies will
occur in increasing order from matching to pure strategies. The
quantitative hypothesis stated that there is no difference between the
predicted and observed stable-state strategies.
In order to test the above mentioned hypotheses, thirty subjects
were randomly chosen from a preschool population. These in turn,
were assigned to the three payoff conditions; no payoff, payoff,
payoff-loss. Each subject was observed under two different experiments
which were identical with respect to all utility relevant features,
but were different with respect to the proportion of the occurrence of the two events, π[subscript i] values. In one experiment π₁ was
set equal to .75 and in the other π₁ was set equal to .80. The apparatus
used was Humphrey's (1939) modified, two-light situation.
This apparatus had never been used before with a preschool population.
The occurrence of the two lights was predetermined randomly
and was held constant during the experiment. The order in which
the two experiments were presented to each of the subjects was
random so that not all subjects received the same π[subscript i]. Reinforcement
was given by the same individual in both the experiments. The
most frequently occurring event was reversed for each of the subjects
from the first to the second testing situation. Under the no
payoff condition the subjects did not receive any extrinsic reinforcement.
Under the payoff condition the subjects received positive
reinforcement only for the correct choice. Under the payoff-loss
condition the subjects were reinforced positively for the correct
choice and reinforced negatively for the incorrect choice. The positive
verbal reinforcement used was, 'That's right, that's very good.'
The negative verbal reinforcement used was, 'That's wrong, that's
not very good.' The data under both the experiments were in the
form of p[superscript -] values, the proportion in which the subject chose the most
frequently occurring event.
Before testing for the overall ordinal hypothesis the data were
tested for sex differences. It was found that there was a significant difference between the stable-state strategies of boys and girls under
the no payoff condition when π₁ = .80 and under the payoff condition
when π₁ = .75.
The ordinal hypothesis which predicted increase in stable-state
strategy with increased payoff was considered untenable as a result
of inspecting the data, the data were such as to suggest that the type
of verbal reinforcement used, namely, a repetitive, simple, monotonous
type, was an ineffective reinforcer. The data were further
examined to see if these subjects had at least a rudimentary grasp of the
concept of probability adequate to the task in the experiment. It
was concluded that the sample of preschool children exhibited little,
if any, understanding of the concept of probability.
In testing the second hypothesis, the Siegel model was used to
quantitatively predict the stable-state strategies of boys and girls
from one experiment to the observed stable-state strategies in the
other experiment. Similarly, the data from the second experiment
was used to predict the stable-state strategies of boys and girls in
the first experiment. It was concluded that the strategies predicted
from the model for both boys or girls under the no payoff and payoff-loss condition did not differ significantly from their observed stable-state strategies. Under the payoff condition, the model predicted
accurately for boys but not for girls.
Heretofore, the Siegel model has been used to predict only with group data. But, since utility refers to subjective value, in this
study the model was used to predict the stable-state strategies of
individual boys and girls, from one condition to the other. It was
concluded that in predicting the individual stable-state strategies the
Siegel model does not predict for girls under conditions of the present