Successful Operant Conditioning of Marine Fish in Their Natural Environment
The cognitive abilities of animals are challenging to assess. Typically, studies on visual discrimination in fishes are conducted in the laboratory neglecting the fact that acclimation of test subjects to the laboratory environment including artificial light conditions may affect the way in which fishes respond to visual stimuli. The aim of this study was to demonstrate that it is possible to train fishes on visual cues using food rewards, while free-roaming in their own territory under natural conditions. An acoustic secondary reinforcer was also included and its potential contribution to the training success was tested. We trained 27 specimens of the benthic triplefin blenny Tripterygion tripteronotum on either a black or white colored target. Training took on average ten days. In 15 binary choice trials, 23 fish discriminated successfully between rewarded target cue (CS+) and the unrewarded cue (CS−). Furthermore, we discuss the usefulness of acoustic secondary reinforcement in this study.
Feeding apparatus and extinction behaviors. (A) Schematic view of the feeding apparatus carrying the two different stimuli, which were gradually moved towards the tube tips as the training progressed. (B) Front view of the black and white stimulus discs the fish were trained on. C and D depict the two behaviors which were defined as disinterest in the stimuli. (C) The fish's entire head is behind the plane in which the stimuli are presented. (D) The fish turns its body axis more than 90° away from the front side of the stimuli.
Duration of individual training phases for all four treatment groups (see Table 1 for sample sizes). Boxplots show median (horizontal line) upper and lower quantiles (boxes) and the range (whiskers). The gray horizontal line shows the overall mean. The training phase was terminated when the individual fish passed the learning criterion to enter the testing phase.
Correct choices of the binary discrimination trials per individual across treatment groups. Every data point represents an individual fish. Right y-axis shows the number of correct choices out of 15 choices. For each fish, the proportion of correct choices (left y-axis) was compared to a 1:1 distribution using a Likelihood Ratio test (P values shown for each ratio). Individuals were ranked by success (x-axis). Within groups of identical success rate, the individuals are ordered according to their treatment groups.
Number of taps until trained tap behavior was extinct. The extinction test was carried out within a single session. Half of the fish were trained with a clicker and received positive reinforcement via the clicker during the extinction test. Neither group received a food reward for tapping.
Contributor Notes
Associate Editor: J. F. Schaefer.