The role of learning and environmental geometry in landmark-based spatial reorientation of fish (Xenotoca eiseni)

Abstract

Disoriented animals and humans use both the environmental geometry and visual landmarks to guide their spatial behavior. Although there is a broad consensus on the use of environmental geometry across various species of vertebrates, the nature of disoriented landmark-use has been greatly debated in the field. In particular, the discrepancy in performance under spontaneous choice conditions (sometimes called "working memory" task) and training over time ("reference memory" task) has raised questions about the taskdependent dissociability of mechanisms underlying the use of landmarks. Until now, this issue has not been directly addressed, due to the inclusion of environmental geometry in most disoriented navigation paradigms. In the present study, therefore, we placed our focus on landmark-based navigation in fish (Xenotoca eiseni), an animal model that has provided fruitful research in spatial reorientation. We began with a test of spontaneous navigation by geometry and landmarks (Experiment 1), showing a preference for the correct corner, even in the absence of reinforced training. We then proceeded to test landmarks without the influence of informative geometry through the use of square environments (Experiment 2-4), varying the numerosity of present landmarks, the distance of landmarks from the target corner, and the type of task (i.e., spontaneous cued memory or reference memory). We found marked differences in landmark-use in the absence of environmental geometry. In the spontaneous memory task, visual landmarks acquired perceptive salience (and attracted the fish) but without serving as a spatial cue to location when they were distal from the target. Across learning in the reference memory task, the fish overcame these effects and gradually improved in their performance, although they were still biased to learn visual landmarks near the target (i.e., as beacons). We discuss these results in relation to the existing literature on dissociable mechanisms of spatial learning.

The geometry as an eyed fish feels it in spontaneous and rewarded spatial reorientation tasks

“…This is an important issue, that emerged here for the first time where we directly compared the two applied procedures. On the other hand, interesting recent results 20 have shown that in the presence of visual local cues at the corners, without the influence of informative geometry through the use of squared environments, the ability to solve the reorientation spatial tasks, by finding the correct corner, remained better in rewarded exit tasks (reference memory) rather than in spontaneous cued memory tasks. When the visual geometric information was present, there was no difference in solution of the task under visual working and reference memory procedures, while in absence of visual geometric information (visual square apparatus), the performance in a rewarded task (reference memory) improved: just as it happened here but under transparent geometric conditions without visual coding.…”

“…For this purpose, two experimental procedures were applied to the behavioural observation of three fish species: a sort of working memory procedure (Experiment 1 and 2), already used with X. eiseni and D. rerio [15][16][17] , based on the spontaneous choices made by animals in a social cued task, and a reference memory procedure (Experiment 3) in a rewarded exit task, applying an operant conditioning with extended learning times, a technique well standardised even with X. eiseni, D. rerio, and C. auratus in similar geometric tasks [9][10][11][12][13][14][18][19][20] .…”

“…For many years now, fishes have been an excellent model to study spatial reorientation in a geometrically characterized environment [9][10][11][12][13][14][15][16][17][18][19][20] . Intriguingly, blind cavefishes have also shown to be able to detect the environmental metric-plus-sense properties, but through extra-visual modalities, such as the lateral line and the sense of touch 21 .…”

“…Differently, the redtail splitfin fish is a viviparous species originating in Mexico and thus inhabits shallow waters with little or no aquatic vegetation. Both zebrafish and goldfish have a large number of neuromasts and their lateral line anatomy is well known 23,31,32 , while the redtail splitfin fish is the prevalent animal model used in visual experiments on environmental geometry's encoding [9][10][11][12]15,18,20 .…”

“…Our behavioural experiments have been conducted in a transparent rectangular arena, by providing three different opportunities to explore the experimental environment. In the present experiment we also used two behavioural procedures: a spontaneous social cued memory task (a sort of working memory procedure [15][16][17]20 ) and a rewarded exit task (a reference memory procedure), similar to those used in other visual geometric tasks with humans and animals, sometimes offering doubts about their comparability [9][10][11][12][13][14][18][19][20] . Interesting was the comparison between these two experimental procedures in visual conditions, since recently divergent results have been shown in fish, when considering the use of local landmarks in non-geometric spatial reorientation tasks 20 .…”