Training and transfer test to study the referential understanding of conspecific photographs by goats

Abstract

Individual recognition requires animals to compare available cues with stored information. For goats, living in stable social groups and forming social hierarchy, it is reasonable to assume they can discriminate between familiar and unfamiliar conspecifics. This study focuses on the cognitive mechanisms underlying goats' perception of conspecific photographs, particularly whether they demonstrate image equivalence. Two groups of goats were trained to discriminate between portrait photographs of familiar and unfamiliar conspecifics. The goats in group A (n = 12) were trained to select familiar individuals, whereas the goats in group B (n = 12) were trained to select unfamiliar individuals. Subsequent transfer test was conducted to assess their ability to generalise learned preferences to novel photographs of previously unseen goats. During the first training tasks (Tr1 and Tr2), no differences in learning performance between the two groups were observed. However, in the later tasks (Tr3 and Tr4), the goats in Group A exhibited better learning performance than did those in Group B. In the transfer test, five goats in Group A, but only one goat in Group B, demonstrated preferences for novel familiar or unfamiliar conspecifics. The superior performance of Group A goats in Tr3 and Tr4 and the number of goats that successfully transferred the familiarity concept to novel individuals provide compelling evidence for the formation of true image equivalence. While goats can establish image equivalence through familiarity, the abstraction of unfamiliar concepts is a more challenging cognitive task.

Keywords: Domestic ungulates; Familiarity; Picture equivalence; Representational insight; Visual discrimination.

Fig. 1

Visual four-choice discrimination tasks used in the last phase of shaping. The rewarded stimulus (S+) in each task is marked with a red circle here for identification purposes only

Fig. 2

Four discrimination tasks (Tr1-Tr4) that were used during the training of experimental Groups A and B. The correct photograph was always of a Group A goat (here, the top left in each case). The other three photographs were of goats from Group B. The four photos in each task are numbered here for identification purposes only (P1-P4)

Fig. 3

Eight new stimulus combinations (Tf1-Tf8) that replaced eight stimulus combinations in Tr4 in the transfer test. The stimulus combinations consisted of one photograph of a goat from Group A that had not been shown before (correct) and three photographs of goats from Group B that had been used before but were rearranged (incorrect). The correct photographs are marked with red circles here for identification purposes only

Fig. 4

Estimated marginal means of individual success rates (%) along with 95% confidence intervals (grey) for Groups A (blue) and B (orange) across the six training days in the four training tasks (Tr1-Tr4)

Fig. 5

Number of trials needed by the animals in Groups A (blue) and B (orange) to reach the predefined learning criterion (TtC) in the four training tasks (Tr1-Tr4). The figure shows the raw data as box plots and the estimated marginal means plus the upper and lower confidence intervals