Are dogs susceptible to serial dependence effects in the relative estimation of quantity?

Serial dependence is a cognitive phenomenon described in humans that alters the perception of a visual stimulus when another similar visual stimulus is previously presented to the observer. Such effect adjusts the perceived appearance of the second stimulus making it look more alike to the first stimulus, therefore producing an attractive perceptual bias (Fischer & Whitney, 2014). The human literature has widely studied the conditions in which serial dependence emerges, and it has been observed that such effect applies to many cognitive aspects such as orientation (Fischer & Whitney, 2014) and numerosity (Corbett et al., 2011). However, the presence of such effect has not been demonstrated in any other species, even though serial dependence could potentially be ecologically advantageous to non-human animals, given that in humans it has been suggested to contribute to the perceptual stability of the surrounding environment (Fornaciai & Park, 2018a). To fill this gap of knowledge, in the current study we examined if serial dependence is present also in dogs, specifically in the cognitive processing of numerosity. We presented 5 dogs with a quantity discrimination task in which they needed to choose the stimulus depicting the larger numerosity between two simultaneously presented stimuli. One of these stimuli was preceded by a task-irrelevant stimulus (i.e., inducer) in its identical location. Three experimental conditions were created: one control condition where no inducer stimulus was presented and other two conditions where either a small or a large inducer stimulus was exhibited. We hypothesized that if dogs were susceptible to serial dependence, the numerosity of the inducer would affect the dogs’ numerical discrimination abilities. Our findings indicate that among the three experimental conditions the dogs’ behavioural responses differed considerably, and such change was statistically represented by a shift of the psychometric curves depicting the dogs’ performances of the non-control trials. When the inducer stimulus was small, the subsequent stimulus presented in the same location (i.e., reference) was underestimated by dogs, making the psychometric curve shift left. Conversely, the curves shifted rightward when the reference stimulus was overestimated, that is when the previous inducer stimulus was large. Hence, the shifts reflected the ability of the inducer stimulus to produce perceptual attractive aftereffects that consequently influenced the dogs’ probability of choosing either quantity. In conclusion, we claimed that numerosity processing in dogs is affected by serial dependence. This result suggests the existence of cognitive perceptual similarities with humans. Ultimately, we speculate that serial dependence might also affect other visual dimensions in dogs and that it could bias the perception of other non-human animals. However, further research on this topic is needed in the future to confirm such speculations.