Effect of age on discrimination learning, reversal learning, and cognitive bias in family dogs

Abstract

Several studies on age-related cognitive decline in dogs involve laboratory dogs and prolonged training. We developed two spatial tasks that required a single 1-h session. We tested 107 medium-large sized dogs: "young" (N=41, aged 2.5-6.5 years) and "old" (N=66, aged 8-14.5 years). Our results indicated that, in a discrimination learning task and in a reversal learning task, young dogs learned significantly faster than the old dogs, indicating that these two tasks could successfully be used to investigate differences in spatial learning between young and old dogs. We also provide two novel findings. First, in the reversal learning, the dogs trained based on the location of stimuli learned faster than the dogs trained based on stimulus characteristics. Most old dogs did not learn the task within our cut-off of 50 trials. Training based on an object's location is therefore more appropriate for reversal learning tasks. Second, the contrast between the response to the positive and negative stimuli was narrower in old dogs, compared to young dogs, during the reversal learning task, as well as the cognitive bias test. This measure favors comparability between tasks and between studies. Following the cognitive bias test, we could not find any indication of differences in the positive and negative expectations between young and old dogs. Taken together, these findings do not support the hypothesis that old dogs have more negative expectations than young dogs and the use of the cognitive bias test in older dogs requires further investigation.

Keywords: Ageing; Cognitive bias; Dog; Learning; Memory; Reversal learning.

Figure 1. Stimuli used in the physical characteristics group.

The two images show the stimuli in the way they are perceived by humans (a) and dogs (b, altered to dog-vision setting through a dedicated image processing tool, http://dog-vision.com). The three plates varied in shape, size and colour. The middle plate (used as ambiguous cue) had intermediate characteristics between the other two plates.

Figure 2. Room set up.

The owner and the dog were at one side of the room (black circles at the bottom of the figure) and the experimenter was standing opposite to them (black circle at the top of the figure). The grey circles in front of the experimenter mark the positions where the plates could be placed based on the type of the stimuli (L1 and L3 = positions for the P and N stimuli in the “location” group; L2 = position for the ambiguous stimulus as well as the P and N stimuli in the “physical characteristics” group).

Figure 3. Survival curve for the number of trials to reach the learning criterion in the discrimination learning according to the age group.

The darker line represents the young dogs, the lighter line represents the old dogs. The shadowed areas reflect the confidence intervals and the crosses (+) indicate censored data (i.e. dogs for which the task was interrupted but they had not learned).

Figure 4. Survival curve for the number of trials to reach the learning criterion in the reversal learning phase of the location group according to the age group.

The darker line represents the young dogs, the lighter line represents the old dogs. The shadowed areas reflect the confidence intervals and the crosses (+) indicate censored data (i.e. dogs for which the task was interrupted but they had not learned).

Figure 5. Survival curve for the number of trials to reach the learning criterion in the reversal learning phase of the physical characteristics group according to the age group.

The darker line represents the young dogs, the lighter line represents the old dogs. The shadowed areas reflect the confidence intervals and the crosses (+) indicate censored data (i.e. dogs for which the task was interrupted but they had not learned).

Figure 6. Box-plot graph for the delta scores of the discrimination learning, consolidation, cognitive bias and reversal learning phases.

The tick bar represents the median score for the group, the boxes are the 25% and 75% quartiles and the error bars represent the minimum and maximum values; * indicates p < .025.