The novel object recognition (NOR) paradigm is a cognitive test that has been used with many species to detect differences in ability. Various iterations of the paradigm have been implemented, making it difficult to compare results both within and across species. Interpretations of the results are equally diverse, threatening the integrity of the paradigm. These inconsistencies have prompted a deeper dive into the variability of the resultant data. For the purposes of this meta-analysis, data originated from 12 studies involving 367 pigs that were subjected to the same NOR paradigm beginning between postnatal days 21 and 24. The main cognitive measure from the NOR paradigm is recognition index (RI), which was the focus of most of the analyses in this meta-analysis. RI was chosen as the main outcome as it determines a pig's preference for novelty, an innate behavior of cognitively intact pigs. A histogram of RI values (range 0 to 1) showed a bimodal distribution skewed to the right, suggesting that the interpretation of positive performance on the task may need to be stricter. Correlational analyses proved that the number of investigations and investigation time with both the novel and familiar objects were the strongest predictors of resultant RI values. Objective data inclusion criteria were then considered to eliminate non-compliant pigs. Results indicated that requiring at least 5 s of investigation over a minimum of 3 investigations with the novel object reduced overall variability for RI with a concomitant increase in the mean. Further analyses showed that pigs preferred to spend more time with and interact more with the novel object across the entire testing trial, especially in the first minute. Together, these findings suggest that future interpretations of NOR should consider applying stricter statistical analyses as well as additional data processing, such as binning, with emphasis on novel object and familiar object investigation. Overall, modifications to the existing iterations of the NOR paradigm are necessary to improve paradigm reliability.
Read full abstract