Fig 1.
Setup for closed-loop activity-dependent protocol.
A–Experimental Setup. The EODs were measured using 8 electrodes, placed on the bottom of the tank (40×30×25) cm. The electrodes were connected to form an array of 5 dipoles: R-1R, R-2R and R-3R, sharing a common reference (R; white circles), A1-A2 (yellow circles) and B1-B2 (red circles). The signal from the 5 dipoles were differently amplified (gain = 50x or 100x: for small fish ~5 cm), summed, squared and then digitized at 25kHz by an ADC board (NI PCI-6521) and stored for posterior analysis. The stimulus pulses were generated by the same ADC board and controlled in real time by a closed-loop real time software that detected the EODs timing and delivered stimulus pulses in response. The stimulus pulses were delivered to the tank by a 7 cm dipole (artificial fish) to mimic an average size of Gnathonemus petersii used in this study. The artificial fish was placed in the middle of the tank as shown in the figure. B–Closed-loop activity-dependent protocol. The real time closed-loop software detected the EODs timing and sent a stimulus pulse after a time delay (d) chosen by the experimenter. The stimulus pulses had the exact Gnathonemus petersii waveform and 3V amplitude to mimic an average size animal. All experiments were performed with the same amplitude and wave shape.
Table 1.
Corresponding fish and delays used in this study.
Fig 2.
Fish response depends on the closed-loop stimulus time delay.
IPI distributions of 4 different experiments are shown for control (dashed lines) and closed-loop stimulation sessions (solid lines), each panel represents the data of a single fish subjected to a single delay for 30 min, the IPIs used to build the histograms were defined in Fig 1B. Top-left–For the stimulus session with time delay d = 10 ms (full line) fish increased the discharges of shorter IPIs, specially around 60 ms IPIs and decreased the probability of firing longer IPIs (> 300 ms) when compared to those of the control session (dashed line). There were also changes in the overall shape of the IPI distribution (50% Pearson's corr., Kolmogorov-Smirnov (KS) p = 0.004). Bottom-left–When stimulated with pulses with time delay d = 12 ms (solid line), fish shortened its IPIs from 5 ms to 200 ms, that is, there was an increase in the frequency of the electric organ as compared to those for the control session (dashed line). For longer IPIs (>200ms) no changes were observed between control and stimulus sessions, i.e., the shape of the IPI distribution remained the same in both sessions with a shift of ~20 ms (52% Pearson's corr., KS p = 0.07). Top-right–For the time delay d = 102 ms (full line), fish increased the probability of firing shorter IPIs of ~20 ms and also longer IPIs (> 300 ms), and decreased the probability of discharging IPIs around 140 ms as compared to those of the control session (dashed line). There were slight changes in the shape of the IPI distribution (89% Pearson's corr. KS p = 0.8). Bottom-right—For stimulus session with time delay d = 172 ms (solid line), fish discharged with high probability 140 ms IPIs and 300 ms IPIs and the shape of the IPI distribution did not change (97% Pearson's corr. KS p = 0.4).
Fig 3.
IPI changes depend on the stimulus time delay.
Quantile-quantile plot of IPIs during control and the closed-loop stimulation sessions for 4 different experiments. The black line represents the reference line y = x, slope = 1. Top-left–the IPIs from the control session and stimulus session with time delay d = 10 ms came from different distributions. Most of the points (blue) were under the reference line indicating that IPIs discharged during the stimulus session were shorter than those of the control session in the range from 50 ms to 375 ms. Bottom-left–The IPIs discharged during the 2 sessions come from similar distribution, IPIs discharged during the stimulus session were ~20ms shorter in a range from 15 to 200 ms. Top-right–IPIs ranging 60 ms to 150 ms came from similar distributions in the control session and in the closed-loop stimulation session. For longer IPIs (>150 ms), the control session presented always shorter IPIs than those of the closed-loop stimulation session, the opposite happened to IPIs shorter than 60 ms. Bottom-right–IPIs from both sessions came from shifted distributions.
Fig 4.
Clustering of IPI changes and analysis as a function of the time delay and the area under the quantile-quantile plot.
If the IPIs discharged during the closed-loop stimulation sessions are shorter than those of the control sessions (top-left), the area under the qqplot curve are smaller than that for the case when IPIs from the 2 sessions come from similar IPI distributions (top-right). The area values for all 22 experiments where individually plotted against the time delays (bottom). The area values were clustered in 2 groups separated by a critical delay d = 100 ms, cluster A (blue crosses; n = 11) containing time delays from 5 ms to 100 ms (except d = 70 ms which was treated as an outlier) and cluster B (red circles; n = 10) from 100 ms to 280 ms. The areas in cluster A were smaller than those of cluster B.
Fig 5.
Global IPI changes differences for 2 clusters.
Mean simple IPI differences between control (X) and stimulus sessions (Y). For the 11 experiments in each cluster in Fig 4, we calculated individually the Tukey mean-difference (Y-X), took the average integrating them in steps of 3 ms and the mean standard deviation (error bars). For cluster A (blue circles), time delays < 100 ms, IPIs discharged during the stimulus sessions were, in average, 90.8 ms shorter than those of the control sessions (blue line), average IPIs from 187 ms to 235 ms presented maximum differences. For cluster B (red triangles), time delays > 100 ms, IPIs discharged during the stimulus sessions were, in average, 3 ms shorter than those of the control sessions (red line). IPIs from 118 ms to 151 ms and from 199 ms to 319 ms were not altered by delays above 100 ms, although IPIs in the last interval had large deviations from the mean. The maximum change in IPIs happened in a range from 157 ms to 196 ms, where fish fired IPIs ~36 ms shorter in the stimulus sessions compared to those in the control sessions. Fish discharged longer IPIs during the stimulus sessions in the IPI ranges from 322 ms to 376 ms. IPIs up to 25 ms were altered independently of time delay.