Contextual effects of small environments on the electric images of objects and their brain evoked responses in weakly electric fish

Alteration of sensory images by their context is a common and important issue in sensory physiology (Palmer,1999). A mirror may change the illumination of a visual scene and the visibility of objects, the resonant cavity of a guitar may amplify the sound generated by the strings, and the presence of minute amounts of spices may transform a regular fare into a delicious dish.

There are theoretical and experimental reasons to believe that contextual effects are also present in electroreception, a sensory modality evolved by aquatic animals (Lissmann,1958; Bullock and Heiligenberg, 1986; Moller,1995). Electric fish are electroreceptive animals that explore their environment with the discharge of specialized electric organs (electric organs, EOs; electric organ discharges, EODs). The EO generates an electric field that is sensed by cutaneous electroreceptors(Lissmann and Machin, 1958; Bullock et al., 1961). Objects in the near environment cause changes in the electric field and in the patterns of transepidermal current density and voltage stimulating the electroreceptive surface. The `electric image' of an object has been defined as the change in the pattern of the transepidermal field caused by that object(Bastian, 1986; Caputi et al., 2002; Budelli et al., 2002).

Two lines of arguments converge to suggest that electric images of given objects are conditioned by the presence of other objects in the electric scene.

First, previous work done by our group has shown that the body of the fish is itself an object that decisively shapes the electric field generated by the EOD (Castelló et al.,2000; Aguilera et al.,2001; Caputi,2004). The fish body implements a pre-receptor mechanism enhancing the reafferent signals at the perioral region, which may be described as an electroreceptive fovea (Castelló et al., 2000; Caputi et al.,2002). The presence of large objects near the fish may alter the above-mentioned funnelling effect, causing consequent changes in the electric image of other objects (Budelli et al.,2002; Rother et al.,2003).

Second, the presence of such large objects is not an exceptional event, but the rule in fish life. In fact, rather than swimming in the middle of the stream, electric fish are frequently found among the roots of floating plants or inside caves on the banks of the rivers. Moreover, most electric fish in captivity choose to stay long periods resting in tubes, or at the angle between the floor and the side of a large object when they are in captivity. These objects may be likened to the structures used as refuges in the nature environment.

This article shows contextual effects on electric images of objects caused by the characteristic electric fish behavior of hovering in tubes. The article also reports the experimental analysis of (i) the physical basis of the changes imposed by small environments on electric images; (ii) the field potential responses at the electrosensory lobe of chronically implanted animals when entering and leaving tubes; and (iii) the effect of context on object discrimination and its implications for electrosensory processing.

We examined contextual effects on electric image generation and processing at the fovea of pulse gymnotid placing the fish outside or inside tubes of different conductivity. Fifteen fish of the genus Gymnotus12–15 cm in length were used. Recent taxonomical studies have revisited the genus Gymnotus and found more than 25 species(Albert et al., 1999); the species we study has been reclassified as Gymnotus inaequilabiatus(Albert et al., 1999). Up to now, however, this species was referred to as Gymnotus carapo (L.)and has been studied by our group for more than 25 years (see the reviews by Caputi, 1999; Caputi et al., 2002). Fish were gathered in the same ecological niche as before, in the Laguna del Sauce lake, Uruguay. Fish collection was performed under the regulations of the Ministry of Ganadería, Agricultura y Pesca. Surgical procedures were performed below 4°C to avoid pain and stress on the fish. At this temperature fish do not react with movements or changes in their EOD rate to nociceptive stimuli. All experiments were done under the rules of the Committee for use of Experimental Animals of the Instituto de Investigaciones Biológicas Clemente Estable and according to the guidelines of the Society for Neuroscience and the International Guiding Principles for Biomedical Research Involving Animals.

1. To measure the time of occurrence of the EOD we recorded the far field on the longitudinal axis of the fish (head-to-tail EOD, htEOD) that yielded the classically described pattern with four wave components generated by different regions of the body, characteristic of the species(V₁–V₄; Trujillo-Cenóz et al.,1984) (Fig. 1A). The largest slope of this htEOD triggered undelayed rectangular pulses (0.1 ms and 5 V) that were used as a trigger for the oscilloscope and acquired through the parallel port of the computer to calculate the inter-EOD interval (signal processing program kindly provided by R. Saa).

2. To evaluate the local potential gradients (sLEOD) we used a probe constructed from two nichrome wires insulated except at their tips (200 μm thick, 2 mm apart; see Castellóetal., 2000). These electrodes were placed along the midline normal to the skin of the jaw. The prefix `s' (for self-generated) is used to refer to the equivalent wave components of the self-generated local field(e.g. sV₁, sV₃, etc.). The effect of resting in tubes on the electric image of a stimulus-object was evaluated using the same experimental procedure described by Aguilera and Caputi(2003) and Caputi et al.(2003). We used a cylindrical stimulus-object (2 mm diameter, 1 cm length) consisting of a plastic tube with its two ends made of graphite carbon discs (1.5 mm in diameter). The cylinder was oriented with its long axis perpendicular to the skin 2 mm away from the electrosensory fovea. When using this stimulus object, the sLEOD was evaluated as the voltage drop between the bare tip of an insulated copper wire (100μm diameter) placed against the skin and the nearest carbon base of the cylindrical stimulus-object. We controlled the longitudinal resistance of the object by connecting an external resistor between both ends of the cylindrical object (Fig. 1B) and measured the voltage drop between the object ends. We used two types of parameters to assess the changes in sLEOD. For global amplitude measurements, the energy of the sLEOD was estimated as the root mean squared value (r.m.s. value) of the sLEOD from an oscilloscope screen trace (10 ms duration, sampled at 100 kHz). For waveform characterization we measured peak voltages (V₂,V₃ and V₄) and calculated their ratios.

3. Electrosensory lobe potentials were recorded by implanting a twisted pair of nichrome wires (insulated except at their tips, 80 μm in diameter),through a small hole in the skull, at depths of 1100 and 1400 μm,respectively. We aimed at the rostral regions of the electrosensory lobe where the foveal region is represented and confirmed electrode position by applying nearby threshold stimuli at the skin. Wires were cemented to the skull using cyanoacrylate and dental cement filler. Recordings were done 1 h after water returned to room temperature when the fish had recovered its natural EOD frequency.

In one case we confirmed the electrode placement histologically. The fish was deeply anaesthetised with ethomidate (2 mg ml^–1) and fixed by aortic perfusion of paraformaldehide (4%). The brain was serially sectioned, mounted and stained with Methylene Blue. The electrode track was clearly visible, with its ending about the centre of curvature of the cell layers of the electrosensory lobe in the molecular layer(Fig. 1C). The cell layers of the electrosensory lobe are shaped like a cup that is concave upward. The currents generated by the whole nucleus will be funnelled to its centre of curvature where the electrode pair was implanted, and the recorded signals can be likened to a holistic view of the activity in the lobe(Lorente de Nó, 1947; Hubbard et al., 1969).

Voltage signals were differences amplified and filtered (band pass 10–10 kHz) using a high input impedance differential amplifier. A digital oscilloscope was used for online observation and averaging of EOD triggered traces (8–64 sweeps). Signals were also recorded for offline measurement and data processing using a Labmaster card and Axotape software.

sLEODs were measured in fish restrained within a band of tissue paper and inside copper or plastic tubes (22 mm inner diameter, 10 cm length) in the presence and in the absence of the stimulus object. Different longitudinal resistances for the object were tested using the procedure described above(Fig. 1B).

The carrier of sensory signals is the basal energy that the object presence modulate for generating images. In the case of active electroreception the carrier is the electric field `illuminating' the object. This field can be considered as generated by an equivalent source completely described by two parameters: its electromotive force and its internal resistance (Thevenin theorem'; Edminister, 1965). In order to estimate these parameters we measured in the same experiments the voltage drop between the two carbon electrodes of the stimulus-object and calculated the current through the object by dividing the measured voltage across the load resistor. The electromotive force and internal resistance correspond to the ordinate intersection point and the slope of the line fitted to the voltage–current plot, respectively.

We hypothesize that the tubes modify the summation of currents generated by different regions of the EO at the foveal region. In order to test this hypothesis, we studied sLEOD components generated by the caudal or rostral portions of the EO when the fish was inside or outside a tube. To dissociate the abdominal and the trunk–tail components of the sLEOD we applied the following procedure. Under cold anesthesia the spinal cord was exposed by a laminectomy and severed between the rostral and middle third of the fish. Spinal section was used to abolish the trunk- and tail-generated EOD (three fish) leaving intact V₁ and a small remnant of V₃generated by the abdominal region(Trujillo-Cenóz et al.,1984; Caputi and Trujillo-Cenóz, 1994). Stimulation of the sectioned cord at the caudal stump was used to evoke the sequence V₂–V₃–V₄ generated by the caudal two thirds of the fish body. Electrical stimuli (0.1 ms, 20 Hz, amplitude supra-maximal for the EOD) were applied through a pair of nichrome wires (200μm thick, 50 kΩ) implanted within the canal to stimulate the bulbo–spinal electromotor tract. In order to assess the completeness of the section and the effectiveness of the spinal cord stimulation we checked the amplitude and waveform of the resultant EOD equivalent electromotive forces using the air-gap technique (see Caputi, 1999).

Increases in stimulus-object conductance and thus image contrast elicited changes in the EOD rate characterized by transient reductions of the inter-EOD interval (novelty responses; Bullock,1969). We used novelty responses to evaluate the fish's electrosensory discrimination ability, as in previous studies(Aguilera and Caputi, 2003; Caputi et al., 2003). Novelty responses evoked by the same changes in object resistance were studied in three experimental conditions: (a) in open field, (b) inside a plastic tube and (c) inside a metal tube (both tubes were 10 cm in length). In these experiments, each trial consisted in the change of image amplitude from a basal (the stimulus-object without resistive load) to a comparison level (set by shunting the carbons for 5 s every 30 s with a known resistor). The sLEOD was simultaneously measured and the change in its amplitude provoked by the variations in object resistance was calculated as the increment of the r.m.s. value (Δ r.m.s.). To detect novelty responses, we plotted the inter-EOD interval sequence. For each response the intervals were numbered starting at the first interval after the resistance change (I₁, I₂....I_n). The baseline inter-EOD interval (I₀) was defined as the mean of the five intervals preceding the change in stimulus-object resistance. We defined the amplitude of the novelty response as the normalized maximum shortening of the inter-EOD interval (novelty response amplitude = 1 – minimum of I_i/I₀, I = 1 to n).

We recorded field potentials in the electrosensory lobe in four fish. To study the dynamic effects of small environments on these field potentials the fish were restrained wrapping them in tissue paper within a long pen made by stretching a plastic netting over a frame consisting of three longitudinal plastic rods (1 cm × 1 cm × 50 cm) held at either end by plastic frames (squared contour of 5 cm × 5 cm). One of the beams was against the bottom of the tank and the others were one at each side. The fish naturally stayed at the bottom of the V-shaped duct and its longitudinal movements were restrained by a couple of cotton balls. Recording procedures for sLEOD and field potentials were as explained above (1 and 2). The relative positions of the electrodes and the fish were visually controlled and remained unchanged during the experiments. Trials in which fish moved were cleared from the records. A U-shaped plastic structure (6 cm between the U arms, 8 cm tall, 10 cm long) was manually moved back and forth with the aid of a system of pulleys (Fig. 1D). One of the pulley wheels was attached to a variable resistor that conducted a constant current square pulse (5 ms, 1 mA) triggered by each EOD. In this way, the recorded voltage drop across the variable resistor coded the structure position at the time of each EOD. We explored the effects of different kinds of longitudinal movement of the plastic structure, including step-like and sine wave-like oscillatory movements of different span,velocities and/or frequencies.

Resting in tubes results in marked changes in the amplitude and waveform of the sLEOD at the foveal region. Fish were recorded in a position that they naturally adopt, in which the fovea region, that is the jaw region(Castelló et al.,2000), is maintained at the tube opening. Plastic tubes increased the r.m.s. value of the LEOD 2.5 times on average (r.m.s. value increase ranged between 1.3 to 3.6 times, N=8; Fischer exact test, P<0.01). Metal tubes had a minor but opposite effect (r.m.s. value decreased between 0.78 and 0.95 times, N=8; Fischer exact test, P<0.01; Fig. 2A). Plastic tubes also cause a large distortion of the ratios of EOD peaks. While sV₂ and sV₄ increase (both in absolute value and relative to V₃), the ratio sV₁/sV₃ decreases. The most dramatic change occurs in sV₄, which increases 4.13 times on average with respect to the control (range 2.97–6.44; Fischer exact test, P<0.01). Besides this absolute increase of V₄,the ratio sV₄/sV₃ increased 1.85±0.09 (mean± s.e.m.; Fig. 2B).

The relative positions of the fish and the tube also change the reafferent stimulus. While the maxima of the r.m.s. value occurred when the snout was just protruding from the limit of the U-shaped plastic structure (in the most commonly observed position adopted by the fish) their minima occurred when the jaw was at the middle of the structure. Important changes in the waveform were also observed as a function of the tube position. Fig. 3 shows sV₃(Fig. 3A) and the ratio sV₄/sV₃ (Fig. 3B) as a function of the distance between of the rostral border of a U-shaped plastic structure and the fish jaw (see Materials and methods, field potentials).

To assess the effects of the tubes on electric images we changed the longitudinal resistance of the cylindrical stimulus-object and recorded the sLEOD at the facing skin (see Aguilera and Caputi, 2003). The sLEOD amplitude decreased as a function of object resistance in all conditions. The sLEOD corresponding to each resistance value of the object changed in proportion to the carrier amplitude,i.e. the amplitude of the sLEOD in the absence of any object (sLEOD increased markedly in the plastic and decreased slightly in the metal tube, Fig. 4A–C).

In Fig. 4, we compared the electric image in three conditions: inside the plastic tube, inside the metal tube or outside the tube (the control condition). While changing the condition caused changes in amplitude and waveform of the sLEOD, the changes in object resistance only affected sLEOD total energy (r.m.s. value) and did not affect the sLEOD waveform in a significant way. This feature and the proportionality between the r.m.s. values of the sLEODs evoked by the same object impedance at the three different conditions explored(Fig. 4C and insets), suggested that the physical mechanism explaining the change in object image is the change in the electromotive force of the electric source `illuminating' the object.

We confirmed this theoretical conclusion in a series of experiments showing that plastic tubes increase the electromotive force of the equivalent electric source `seen' from the object in an amount similar to the increase of the carrier and electric image. In Fig. 5 voltage measured between the object tips is plotted as a function of the current through the object. This allowed us to calculate the electromotive force (ordinate intersection point) and the series resistance(slope) of an equivalent source `seen' by the object. This indicates that plastic tubes increase object's electric `illumination' which, in turn,increase the imprimence (see Discussion) that they cause and also increase the constrast of the electric image that they project. The plot of Fig. 5A shows the r.m.s. value of the voltage between the poles of the stimulus object as a function of the current through the object. While the slope of the line, corresponding to the internal resistance of the electric source `illuminating' the object was constant, the ordinate intersection, (i.e. the electromotive force) increased by 2.5 times (the same amount as the increase in LEOD).

The change in waveform of the sLEOD also reflected the change in waveform of the electromotive force of the `illuminating' source. Wave components generated more caudally showed a more pronounced increase. While sV<sub>3</sub> generated all along the fish exhibits the smaller increase(about 2 times), sV<sub>4</sub> generated mainly at the tail exhibits the maximum increase (about 3.5 times); sV<sub>2</sub>, generated at the centre of the fish body increases in an intermediate manner (about 2.5 times). This is shown in Fig. 5B–D by the gradation of the increase in ordinate intersection. Taking into account these results and that the sLEOD is the `weighted sum' of multiple regional EOD components, we hypothesized that caudally generated currents reach the foveal region in larger proportion when the fish is inside a plastic tube.

This hypothesis was tested by altering the electrogenic properties of different regions of the EO in animals by spinal sections. In three fish the sLEOD was recorded inside and outside the tube before and after spinal section silencing the caudal region of the EO. When the spinal cord was severed at about one third of the distance from head to tail, the remaining EOD consisted mainly of sV₁ (Fig. 6A, orange trace). In addition, when stimulating the distal stump of the severed spinal cord we recruited the electromotoneuron pools and reproduced the normal activation of the caudal region of the EO generating sV₂, sV₃ and sV₄(Aguilera, 1997; Caputi, 1999; Fig. 6B, orange trace). When the fish was inside the tube, sV₁ decreased to about one half of the control value (Fig. 6A,green trace) and the caudally evoked sV₃ increased about 3 times the control value (Fig. 6B,green trace). We concluded that the carrier changes are explained by the simple physical model described in Fig. 6C,D.

A first strategy to assess how the changes in the electrosensory carrier and images are relevant for sensory processing was to explore discrimination of small resistive objects facing the fovea. In six fish, we applied the technique described by Aguilera and Caputi(2003). Fig. 7A–C show the novelty responses when changes in object conductance from open circuit to 4.5μS, 15 μS and 2000 μS were applied respectively. When fish were inside plastic tubes the amplitude of the novelty response was significantly larger than seen in the control (Fischer exact test, P<0.01; 6 fish and 6 different changes in object conductance). No significant differences were found between the metal tube and control experiments. This implies that there is a larger sensitivity of the electrosensory system to the same change in object conductance when the fish is in the plastic tube. Moreover, changes in stimulus-object conductance that were close to threshold for eliciting the novelty response in the metal tube or in control conditions,always generated a novelty response when the fish was inside the plastic tube. The amplitude of the novelty response was an increasing function of the change in object conductance and the presence of the tube shifted the curve upward and the abscissa intersection point (threshold) to the left(Fig. 7D), indicating an increase in sensitivity for the same change in object conductance. However,when the amplitude of the novelty response was plotted as a function of the change in sLEOD amplitude (ΔsLEOD), a single logarithmic function fit the data obtained in both conditions (Fig. 7E), which suggests that the ability to discriminate changes in object image amplitude was not altered by the presence of the plastic tube.

A second strategy to assess how the changes in the electrosensory carrier and images are relevant for sensory processing was to record field potentials in animals chronically implanted with electrodes in the rostral regions of the ELL where the electrosensory fovea is represented.

Field potentials showed a characteristic pattern of response following the EOD (Fig. 8A). These responses show three clear components that can be assigned to the fast and slow electrosensory pathways defined by Szabo et al. (1973): (i) A brief spike occurs at short latency (1–3 ms) after the EOD. This corresponds to the fast electrosensory pathway and will be referred to as FEP response(Castelló et al.,1998). (ii) An early slow wave starting at about 2–4 ms after the EOD and lasting about 5 ms. In some recordings this response showed small spikes, probably corresponding to the synchronized activity of primary afferents and/or granule cells from the deeper layers of the ELL (referred to as SEP early response). (iii) A series of slow waves starting about 7–10 ms and lasting for the rest of the interval between EODs (SEP late response). This response showed larger variability than the early SEP and probably corresponds to the activity of the more dorsal layers of the ELL(Maler, 1979).

To confirm that these three are evoked electrosensory responses we studied their variability when the sLEOD amplitude was altered by plastic U-shaped structure movements. We calculated the peri-EOD standard deviation across an ensemble of 300 subsequent epochs of 30 msduration, each epoch starting 5 ms prior to the EOD. We plotted the standard deviation of the voltages recorded at a given time after the EOD as a function of such time(Fig. 8B) We compared six sets of responses each obtained using a different experimental protocol altering the sLEOD (top plot, black traces). As a reference the post-EOD averaged response when the fish was in an open field condition is displayed in the bottom plot. This analysis shows three clear peaks in the standard deviation of the signal that we assigned to the changes in the activities of the fast electrosensory pathway (FEP) and the early and late components of the slow electrosensory pathway (SEP). We tested the alternative hypotheses that these changes in the response were provoked by changes in the inter-EOD interval or by lateral line stimuli. In the same fish we recorded the ELL evoked responses by a nearly constant sLEOD when the fish was stimulated by vibratory stimuli causing large EOD accelerations. We recorded an ensemble of 300 consecutive peri-EOD epochs obtained with the fish resting in open field while gentle taps on the aquarium wall were applied. The post-EOD standard deviation did not produce the characteristic trimodal profile observed when the sLEOD was modified. Instead it showed a flat profile(Fig. 8B, red trace),confirming that the three peaks coincident in time with the three described components of the field potentials are caused by changes in electrosensory signals but not by vibratory stimuli or by the change in EOD interval.

Discrimination experiments suggested the presence of adaptive mechanisms to avoid saturation of the electrosensory system. We studied whether the different responses in the electrosensory lobe showed adaptation when fish goes in and out small environments.

The FEP response consists of a spike. The amplitude of this spike is linearly correlated with the r.m.s. value of the preceding pulse(Fig. 8C). The slow electrosensory pathway (SEP) response was characterized by two main components characterized by their latency and dynamics. The early SEP response showed little adaptation, whereas the late response showed clear adaptation. Fig. 9A shows the electrosensory lobe field potentials (color coded) in response to a sequence of sLEODs when the tube was moved along the fish axis coming from the caudal region. In this colormap the horizontal dimension corresponds to time after the EOD and the vertical dimension to the course of the experiment. Both early and late potentials changed with tube position. To analyse the change in these responses we subtracted the average response in open field from every evoked response (results shown as a color map in Fig. 9B). We measured the amplitude of the response at selected latencies corresponding to the early and late SEP responses (vertical lines). While the amplitude of the early response remained similar after the step in the sLEOD, the late response showed a progressive attenuation (Fig. 9C,D). In addition, we observed in all fish that sudden increases or decreases of the sLEOD provoked by step-like movements of the tube evoked very large responses at about 7–10 ms after the EOD. This transient component increased in latency and diminished in amplitude, disappearing in less than 1 s (10–15 EODs). Fig. 9E shows three consecutive evoked responses just after a step-like movement of the tube (red traces) caused a change in sLEOD amplitude. These traces are compared with the averaged field potentials obtained outside the tube (green traces) and inside the tube, at the best position (blue traces). In one of the four fish this transient response to a step change in sLEOD appeared clearly separated in time (Fig. 10, single vertical arrow) from the rest of the late response,which exhibited a much less pronounced adaptation(Fig. 10, double vertical arrows).

The present study provides a clear example of how the environmental context introduces changes in object representation at the sensory surface and in the central nervous system. This effect (comparable in vision to painting the walls black in the case of metal tubes or covering them with mirrors in the case of the plastic tubes) is generated by alteration of the image generation post-effector/pre-receptor mechanisms. This effect is not an experimental artefact since it occurs in natural conditions when fish choose to rest between roots or inside caves. Interestingly, even though the baseline stimulus is more than doubled, the fish react with similar novelty responses to similar changes in sLEOD when the fish is inside a plastic tube. This suggested the presence of adaptive mechanisms that were confirmed by the field potential responses recorded in the ELL.

This discussion address three important points about contextual effects of resting in a tube: (a) the physical mechanisms underlying changes in object image caused by small environments, (b) how these contextual effects affect sensory discrimination and (c) how the electrosensory lobe deals with the large changes in waveform and amplitude of object images caused by small environment contextual effects.

Lissman and Machin (1958) coined the expression `object imprimence' to describe the `imprint' of an object on the fish's self-generated electric field. Imprimence may be viewed as a virtual electric field caused by the presence of the object (Lissman and Machin, 1958). Since the `electric image'of a given object is defined as the change in the pattern of the transepidermal field caused by the presence of the object, such an `electric image' is just the object imprimance at the sensory surface.

All materials other than water cause electric imprimence on the EO generated field. Because the imprimance of an object is also an electric field, the imprimance of every object is modified by the presence of other objects and reciprocally. Therefore, the electric image of a given object is highly dependent of the presence of other objects(Rother et al., 2003).

The fish body is the object that most decisively shapes the electric field generated by the EOD and consequently the electric imprimance of surrounding objects. This is because the low impedance of the internal tissues and the silhouette of the fish body facilitates the flow of current from caudal to rostral regions of the fish(Castelló et al., 2000; Aguilera et al., 2001; A. Migliaro, A. A. Caputi and R. Budelli, unpublished data). In addition, the caudal region behaves as a `current generator', forcing current rostrally. Therefore, the carrier and the local reafferent signals are enhanced in the perioral region. In addition, the number and variety of receptors as well as their central projection field are maximal in this region. For these reasons the sensory mosaic at the perioral region has maximal spatial resolution and has been likened to an electrosensory fovea(Castelló et al.,2000). Our experiments confirm the hypothesis that a tube surrounding the fish body modulates such funneling effect and modifies electric images of nearby objects at the fovea.

We showed that the plastic tube decreased the lateral shunting of current,facilitating funnelling to the perioral region of the faster waves(sV₂–sV₃–sV₄) generated at the caudal region of the EO. Thus, the most caudally generated sV₄ is the component that increases the most. In addition, the slow early component(sV₁) as well as the abdominally generated component of the positive wave (sV₃) are reduced because abdominally generated current must flow along the tube, facing a larger resistance path. Since the abdominal region of the EO acts as a `voltage' source, the generated current decays because of the voltage drop that occurs along a path of higher resistance (Fig. 7C,D).

The presence of a metal tube causes a different effect. It short-circuits the return of the currents generated at the caudal regions, preventing their funneling to the fovea. Nevertheless, the low internal resistance of the`voltage' equivalent source at the abdominal region of the EO allows it to maintain most of its contribution to the sLEOD. For this reason, the decrease of the sLEOD inside a metal tube is not as dramatic as its increase inside a plastic tube.

One of the most important functions of sensory systems is to highlight perceptual experience of the attributes of objects that are more closely correlated with the intrinsic properties of the object than with their images. The objects in the environment are called distal stimulus while the image on the sensory mosaic is called the proximal stimulus(Palmer, 1999).

This function of the nervous system allows an evaluation of the attributes of objects independently of the context or scene in which they are immersed. For example, the whiteness of a paper remains the same whether looked at indoors or outdoors under the sunlight (see also Adelson's web page web.mit.edu/persci/people/adelson/checkershadow_illusion.html). This is not the case for fish evaluation of electric attributes of objects that lead to novelty responses. In our previous studies we showed that novelty responses were graded with both the change in object impedance and the change in their corresponding images (sLEOD; Caputi et al., 2003). However,in those experiments the changes in sLEOD were correlated one-to-one with the changes in object impedance, and therefore it was impossible to answer the question about which of those variables determines the amplitude and threshold of the novelty response. The increase in carrier amplitude and object image generated by plastic tubes allowed us to compare the effects of the same change in object impedance when the change in image amplitude was different. It also allowed us to compare the effects of similar changes in image amplitude when the change in object impedance was different. The experiments reported in the present article show that fish respond with larger novelty responses to the same change in object resistance when the change in image was larger. In addition, similar changes in the r.m.s. value of the sLEOD generated similar novelty responses. Thus, novelty response depends only on changes of the object's image. Thus, the novelty response depends on the proximal stimulus at the receptor surface, not on the distal stimulus in the environment. It should be noted, however, that at distal stimulus perception(i.e. evaluation of the absolute attributes of the object) probably occurs in parallel by the central nervous system of these fish. The presence or absence of such function must be evaluated, therefore, by indicators other than the novelty response.

The discrimination experiments reported here suggest the presence of central adaptive mechanisms in the sensory evaluation of electric images. Even though the basal sLEOD in plastic tubes is about 2.5 times the basal stimuli observed in open field, and its waveform is different, the amplitudes of novelty responses generated by similar energy changes in the sLEOD were similar. This indicates that the increase of the basal image caused by the plastic tubes does not saturate the sensory system, suggesting that adaptation takes place at central structures. If receptors were adapting to the basal stimuli they would probably evaluate the change in stimuli differently, which in fact does not occur. In addition, this confirms that the basal sLEOD waveform and amplitude are independently subtracted from the present image to detect novelty (Aguilera and Caputi,2003; Caputi et al.,2003).

The fast electrosensory pathway (FEP) was easily identified by the characteristic early, large peak that was modulated in latency and amplitude,without adaptation, by changes in the amplitude of the sLEOD(Castelló et al.,1998). The simplicity of the fast electrosensory response corresponds to the simplicity of its circuitry, and the complexity of the slow electrosensory response (variability, multiple components and long duration)corresponds to the complexity of its circuitry(Carr and Maler, 1986). Analysis of the standard deviation of the voltage recorded at a given time after the EOD indicates the presence of an early and a late stage in the SEP response. The latter could be broken down into subcomponents of different dynamics in relation to the sequence of images.

Timing and dynamics of the early slow responses suggest that they may correspond to the deepest layers of the ELL, including primary afferents and cells of the granule layer (Maler,1979; Carr and Maler,1986; Bastian,1986). The late slow electrosensory responses observed in this study may correspond to more dorsal layers of the electrosensory lobe including the polymorphic cell layer, and the effects of recurrent inputs coming from higher centers (Bastian,1995; Berman and Maler,1999). These late responses show adaptation with two different rates of decay after step and hold stimuli. Step-like changes in the electrosensory image caused the appearance of a transient wave at the beginning of the late response that rapidly decayed and increased in latency. The step-like increases of the sLEOD also evoked a long-lasting change of the late response that exhibits a slow decay. Interestingly, the presence of two subcomponents having different time constants may suggest the presence of different types of responses in the cells of the ELL. The different dynamics of these responses is not surprising since the presence of adapting and non-adapting cells differently involved in the recurrent electrosensory loops was recently described in wave gymnotids(Bastian et al., 2004). These adaptive responses in the ELL of pulse gymnotids appear similar to those described in the electrosensory systems of other fish(Bell et al., 1993; Bastian, 1995; Bastian et al., 2004; Bodznick et al., 1999). Our study suggests that the discrimination rules described in G. carapoby Caputi et al. (2003) may partially be implemented at the electrosensory lobe by the storage and subtraction of a moving average of past electric images from the current input.

1. Hovering in small environments is a common behaviour of fish. We show that this behaviour might cause large contextual effects on electric imaging.

2. These effects are due to changes in efficacy of a previously described pre-receptor/post-effector funnelling effect based on the high conductivity of the fish body (Castelló et al.,2000). Because current funnelling efficacy is altered by the presence of a tube, the electromotive force of the equivalent source electrically `illuminating' the object is modified in amplitude and waveform,with the consequent change in object image on the skin.

3. The study of object discrimination under two different contextual conditions indicates that fish evaluate the changes in images (proximal stimulus) rather than changes in object attributes (distal stimulus), and suggest the presence of adapting mechanisms at the central nervous system dealing with background subtraction.

4. Electrosensory lobe field potentials recorded in chronically implanted fish in and out of small environments confirm the presence of adapting mechanisms at the slowelectrosensory pathway, and also show the presence of non-adaptive responses in both the fast and slow electrosensory pathways.

We thank to J. A. Ferreira for technical assistance during the experiments and to Dr C. C. Bell for critical comments on an early version of this manuscript. Partially supported by Fogarty grant number 1R03-TW05680, ECHOS Project U03B01 and PEDECIBA.