A new human delayed‐matching‐to‐place test in a virtual environment reverse‐translated from the rodent watermaze paradigm: Characterization of performance measures and sex differences

Abstract

Watermaze tests of place learning and memory in rodents and corresponding reverse‐translated human paradigms in real or virtual environments are key tools to study hippocampal function. In common variants, the animal or human participant has to find a hidden goal that remains in the same place over many trials, allowing for incremental learning of the place with reference to distal cues surrounding the circular, featureless maze. Although the hippocampus is involved in incremental place learning, rodent studies have shown that the delayed‐matching‐to‐place (DMP) watermaze test is a more sensitive assay of hippocampal function. On the DMP test, the goal location changes every four trials, requiring the rapid updating of place memory. Here, we developed a virtual DMP test reverse‐translated from the rat watermaze DMP paradigm. In two replications, participants showed 1‐trial place learning, evidenced by marked latency and path length savings between Trials 1 and 2 to the same goal location, and by search preference for the vicinity of the goal when Trial 2 was run as probe trial (during which the goal was removed). The performance was remarkably similar to rats' performance on the watermaze DMP test. In both replications, male participants showed greater savings and search preferences compared to female participants. Male participants also showed better mental rotation performance, although mental rotation scores did not consistently correlate with DMP performance measures, pointing to distinct neurocognitive mechanisms. The remarkable similarity between rodent and human DMP performance suggests similar underlying neuro‐psychological mechanisms, including hippocampus dependence. The new virtual DMP test may, therefore, provide a sensitive tool to probe human hippocampal function.

Reinforcement Learning approaches to hippocampus-dependent flexible spatial navigation

“…However, when the goal location changes, the model needs many trials to adjust in order to accurately navigate to the new goal, which is in marked contrast with real DMP performance of rats ( Fig. 1b) and humans (Buckley and Bast 2018).…”

“…Hence, the model can be considered as a model of one shot recall, rather than one-shot learning. This cannot account for the onetrial place learning performance shown by rats and human participants on DMP tasks towards new goal locations, rather than familiar ones (Bast et al 2009;Buckley and Bast 2018). However, the hierarchical RL model may account for onetrial place learning performance on the DMP task when the changing goal locations are familiar goal locations, i.e.…”

“…b. and c. in Bast et al (2009) for search preference across days) and human participants (Buckley and Bast 2018). Moreover, the random search during trial 1 is inconsistent with the finding that rats on the DMP task (but not human participants, (Buckley and Bast 2018)) tend to go towards the previous goal location on trial 1 to a new goal location (Steele and Morris (1999); Pearce et al (1998), and our own unpublished observations); in addition, both rats and human participants show systematic search patterns on trial 1 (Buckley and Bast 2018;Gehring et al 2015). The adjustment of the policy when the predicted goal is not encountered is not addressed in the current approach, a point that section 4.3 covers.…”

“…1c) with the platform removed (Bast et al 2009). Buckley and Bast (2018) reverse-translated the watermaze DMP task into a task for human participants, using a virtual environment presented on a computer screen, and have shown that human participants show similar one-trial place learning to rats.…”

“…Such one-trial place learning is reflected by a marked latency reduction between the first and second trials to a new goal location (Fig.1b), with little further improvement on subsequent trials, and by a marked search preference for the vicinity of the correct location when trial 2 is run as probe (Fig.1c) with the platform removed (Bast et al 2009). Buckley and Bast (2018) reverse-translated the watermaze DMP task into a task for human participants, using a virtual environment presented on a computer screen, and have shown that human participants exhibit similar one-trial place learning to rats.…”