Predictive OKR during Training in Different Fish Species. Eye movement performance of a representative normal goldfish (A), carp (B), zebrafish (C), medaka (D) and vestibular neurectomized goldfish (E) at the beginning (left, Control) and the end (middle, Trained) of 30-min bidirectional velocity step training with half-stimulus period of 8 sec. Right panels show mean eye velocity traces of each fish group (normal goldfish n = 13, carp n = 3, zebrafish n = 9, medaka n = 13, neurectomized goldfish n = 7) averaged over the initial (Control, pale color) and the last (Trained, dark color) 18 stimulus cycles (for 5 min). Numbers by the arrows indicate mean Predictive Deceleration values.
Predictive OKR during Extended Period and in Dark in Different Fish Species. Eye movement performance during extended period stimulation with a half-period of 16 sec (Left) and in the dark (Right) after bidirectional velocity step training whose half-period was 8 sec in representative goldfish (A), carp (B), zebrafish (C), and medaka (D).
OKAN Habituation after Acquisition of Predictive OKR in Goldfish. (A) OKAN of a representative goldfish after 1st (top) and 5th (middle) constant velocity visual stimulation for 1 min. The bottom trace shows OKAN of the same goldfish after acute cerebellectomy after the 5th test. (B) OKAN tested before (top) and after (bottom) unidirectional (clock-wise) velocity step training
OKAN in Different Fish Species. OKAN of a representative normal goldfish (A), carp (B), zebrafish (C), medaka (D), and neurectomized goldfish (E) along with mean OKAN duration in each fish species (right). Error bars are standard deviation. OKAN durations of goldfish (both normal and neurectomized) and carp are time constant τ [s] in Eq. (
OKAN duration vs. Acquired Predictive Deceleration of individual fish. Orange: normal goldfish (n = 13), red: carp (n = 3), brown: neurectomized goldfish (n = 7). Predictive Decelerations were measured from averaged eye velocity traces over the last 25 stimulus periods of 3-h bidirectional visual training. OKANs were measured before visual training, and time constants τ were estimated by using Eq. (
Effects of Vestibular Neurectomy in Goldfish. (A) Schematics of right semi-circular canals and sectioned vestibular nerves. Vestibular neurectomy was done bilaterally. (B) Horizontal vestibuloocular reflex measured in the dark in response to a sinusoidal head velocity stimulation at 0.125 Hz with the amplitude of 40 deg/s. Thin cyan lines indicate eye velocity traces during each of 30 stimulus cycles, and thick blue lines indicates their average. (C) Comparison of averaged eye velocity traces just after (Control) and 3 hours after (Trained) the beginning of bilateral velocity step visual training (gray) between normal (pale and dark orange traces) and vestibular neurectomized goldfish (pale and dark brown traces). (D) Learning curves of predictive deceleration of normal (orange) and vestibular neurectomized goldfish (brown). Predictive decelerations (a in Eq. (1) in Methods) were estimated from averaged eye velocity traces over 13 fish (normal) and 7 fish (neurectomized) for each stimulus cycle. A moving average spanning 11 stimulus cycles was applied to both learning curves to reduce variabilities in the estimation of predictive deceleration.
Figure 5E illustrates the OKAN of a typical neurectomized goldfish measured before visual training in comparison with normal goldfish (A) and other fish species (B-D). In contrast to intact goldfish (A), the OKAN of the representative neurectomized goldfish lasted significantly shorter as previously demonstrated in goldfish36 and other animal species32,34. Mean OKAN duration for 7 neurectomized goldfish was 4.5 s (Fig. 5E, right panel) in contrast to 32.2 s for normal animals. The variability of OKAN duration time in neurectomized animals (SD: 1.2 s) was much smaller than that in the normal group (SD: 25 s).
Figure 2E illustrates a typical horizontal eye velocity trace of a vestibular neurectomized goldfish recorded just after (Control, left panel) and 30 min after (Trained, middle panel) the beginning of the same visual stimulation (gray) as used for other fish in Fig. 2. Neurectomized goldfish generated seemingly normal OKR eye velocity with greater Initial Jump and faster Build-up than the normal animals (A, left panel). After 30 min of visual training (E, Trained, middle panel), eye velocity started to decrease before changes in the stimulus direction. This was also clearly seen in the averaged eye velocity traces over 7 neurectomized fish (E, right panel) in Control (right brown line) and in Trained (dark brown line). The averaged Trained eye velocity clearly started to slow down before changes in the stimulus direction (dark brown arrows). Predictive Deceleration values for Control and Trained are −0.13 deg/s2 and −0.94 deg/s2, respectively, and the difference is statistically significant (t-test, p = 0.0041).
In order to further evaluate differences in predictive OKR between normal and neurectomized animals in a fully adapted state, we compared in Fig. 7C averaged eye velocity traces at the beginning (Control) and 3 h after the visual training (Trained). Averaged eye velocity of neurectomized animal (Control, light brown) arose (Initial Jump) more rapidly than that of normal animal, and reached a plateau level slightly slower than the maximum velocity of the normal animals in both CW and CCW stimulus rotations. After 3 h of visual training, Trained eye velocity of normal animals arose more rapidly (greater Initial Jump), reached a greater maximum velocity than Control eye velocity, and began to slow down before changes in the visual stimulus direction as described in Figs. 1B, 2A and a previous study3. Similarly, Trained eye velocity of neurectomized animal jumped-up more rapidly, reached a greater maximum velocity than Control, and slowed down prior to changes in the visual stimulus directions. Namely, neurectomized goldfish acquired predictive OKR as in normal animals although on average Predictive Deceleration appeared smaller than in normal fish.
Figure 7D compares the averaged time courses of Predictive Deceleration for neurectomized goldfish (brown, n = 7) and normal animals (orange, n = 13). The orange trace for normal fish started from a positive value at the beginning of the visual training (Control) and decreased to negative, eventually reaching a plateau around −3 deg/s2 after 3 h. By contrast, the time course for neurectomized animals started near 0 and gradually decreased more negatively at a slower rate than the normal animals. Eventually a plateau was reached around −2 deg/s2 after 3 h.
To compare the relationship between OKAN duration and acquired Predictive Deceleration with that of normal goldfish, data from each of 7 neurectomized goldfish were superimposed in Fig. 6 (brown filled triangles). As noted above, OKAN durations were significantly shorter and less variable than those of normal animals with data presented on the left side along the abscissa. The neurectomized data are scattered in the region between 1 and 3 deg/s2 on the ordinate (horizontal dotted lines) in which reside data from 9/13 normal goldfish and 2/3 carp. These plots showed that neurectomized goldfish with OKAN duration severally shortened acquired predictive OKR comparable to about 70% of the normal goldfish population.