Effects of proactive interference on olfactory memory in dogs.

Proactive interference (PI) occurs when memories of past events or stimuli intrude in the present moment, causing working memory (WM) errors. These errors are often measured through WM tests such as matching-to-sample (MTS). When the repetition of individual stimuli increases, there is a greater chance of these intrusions, and thus there can be a decrease in accuracy in such tasks. In two experiments, we explored the nature of PI on dog working memory. First, we manipulated the size of the set of odors (2, 6, trial-unique) used to construct each session to maximize (2-odor set) and minimize (trial-unique) within-session proactive interference during an olfactory MTS task. Matching-to-sample accuracy decreased with greater PI. Second, we adapted procedures originally designed for pigeons and rhesus macaques to determine the locus of PI in dogs. To test for proactive interference, probe trials were inserted into MTS sessions where sample odors from earlier trials reappeared as incorrect comparisons. Incorrect responses on these probe trials indicated proactive interference. These probe tests were conducted with a 0-s or 20-s retention interval in separate sessions. We found that dogs performed worse on the matching task when the source of interference (odor stimulus) was from the immediately preceding trial compared with when they were from trials further back in the session but only for the 0-s retention interval. These results are compared with previous work examining the effects of proactive interference on working memory in other species.

Dissociating the effects of delay and interference on dog (Canis familiaris) working memory.

Delayed matching-to-sample (dMTS) is commonly used to study working memory (WM) processes in non-humans. Previous procedures for studying dog WM, including versions of the dMTS, did not separate the impact of delay and interference on memory performance. These studies were also limited to auditory and spatial stimuli, neglecting dogs' dominant sensory modality (i.e., olfaction). Therefore, we designed the first olfactory dMTS in dogs, with systematically varied delays and number of odors in a session, to dissociate the effects of delay and within-session proactive interference on dog WM. Dogs (n = 5) initially trained on matching-to-sample with 48 odors, with a zero-second delay, were tested on four delay lengths (0, 30, 60, and 90 s), counterbalanced across three, trial-unique, sessions. Although there was a slight decrease in accuracy across delays, dogs performed above chance on delays up to 60 s, suggesting a WM duration of at least 60 s. To explore the effect of within-session proactive interference on WM duration, the size of the stimulus set was reduced to six and two odors. There was no effect on the memory function with six odors compared to the trial-unique sessions. However, the interference caused by the two-odor set was enough to decrease accuracy at each delay length. These findings suggest that forgetting in dog working memory for odors can be simultaneously influenced by delay and within-session proactive interference.

Matching-to-sample abstract-concept learning by dogs (Canis familiaris).

The abstract concept of sameness forms the basis of higher-order cognitive operations, including mathematics and language. Historically believed to be unique to humans, evidence of abstract-concept learning in recent decades has been demonstrated in a range of phylogenetically diverse species, indicating that the ability to judge sameness relations is a general process resulting from convergent evolution. However, to date, no research has demonstrated evidence of such learning in any canid species. We trained domestic dogs (n = 6) on a two-choice olfactory matching-to-sample task using a training set of 48 odors in trial-unique sessions. Upon meeting an acquisition criterion (two consecutive sessions ≥ 83% correct), we assessed abstract-concept learning by testing for transfer to novel odors. Dogs matched novel odors with above-chance accuracy and exceeded baseline levels, satisfying previously proposed criteria for full abstract-concept learning. Our findings provide the first evidence of MTS concept learning in dogs, illustrating qualitative similarities with other species. (PsycInfo Database Record (c) 2021 APA, all rights reserved).