Encoding and processing biologically relevant temporal information in electrosensory systems.

Wave-type weakly electric fish are specialists in time-domain processing: behaviors in these animals are often tightly correlated with the temporal structure of electrosensory signals. Behavioral responses in these fish can be dependent on differences in the temporal structure of electrosensory signals alone. This feature has facilitated the study of temporal codes and processing in central nervous system circuits of these animals. The temporal encoding and mechanisms used to transform temporal codes in the brain have been identified and characterized in several species, including South American gymnotid species and in the African mormyrid genus Gymnarchus. These distantly related groups use similar strategies for neural computations of information on the order of microseconds, milliseconds, and seconds. Here, we describe a suite of mechanisms for behaviorally relevant computations of temporal information that have been elucidated in these systems. These results show the critical role that behavioral experiments continue to have in the study of the neural control of behavior and its evolution.

Interval-integration underlies amplitude modulation band-suppression selectivity in the anuran midbrain.

We examined the mechanisms that underlie 'band-suppression' amplitude modulation selectivity in the auditory midbrain of anurans. Band-suppression neurons respond well to low (5-10 Hz) and high (>70 Hz) rates of sinusoidal amplitude modulation, but poorly, if at all, to intermediate rates. The effectiveness of slow rates of sinusoidal amplitude modulation is due to the long duration of individual 'pulses'; short-duration pulses (<10 ms) failed to elicit spikes when presented at 5-10 pulses s(-1). Each unit responded only after a threshold number of pulses (median=3, range=2-5) were delivered at an optimal rate. The salient stimulus feature was the number of consecutive interpulse intervals that were within a cell-specific tolerance. This interval-integrating process could be reset by a single long interval, even if preceded by a suprathreshold number of intervals. These findings indicate that band-suppression units are a subset of interval-integrating neurons. Band-suppression neurons differed from band-pass interval-integrating cells in having lower interval-number thresholds and broader interval tolerance. We suggest that these properties increase the probability of a postsynaptic spike, given a particular temporal pattern of afferent action potentials in response to long-duration pulses, i.e., predispose them to respond to slow rates of amplitude modulation. Modeling evidence is provided that supports this conclusion.

Short-term synaptic plasticity as a temporal filter.

Synaptic efficacy can increase (synaptic facilitation) or decrease (synaptic depression) markedly within milliseconds after the onset of specific temporal patterns of activity. Recent evidence suggests that short-term synaptic depression contributes to low-pass temporal filtering, and can account for a well-known paradox - many low-pass neurons respond vigorously to transients and the onsets of high temporal-frequency stimuli. The use of depression for low-pass filtering, however, is itself a paradox; depression induced by ongoing high-temporal frequency stimuli could preclude desired responses to low-temporal frequency information. This problem can be circumvented, however, by activation of short-term synaptic facilitation that maintains responses to low-temporal frequency information. Such short-term plasticity might also contribute to spatio-temporal processing.

Short-term synaptic plasticity contributes to the temporal filtering of electrosensory information.

Short-term synaptic depression and facilitation often are elicited by different temporal patterns of activity. Short-term plasticity may contribute, therefore, to temporal filtering by impeding synaptic transmission for some temporal patterns of activity and facilitating transmission for other patterns. We examined this hypothesis by investigating whether short-term plasticity contributes to the temporal filtering properties of midbrain electrosensory neurons. Postsynaptic potentials were recorded in response to sensory stimuli and to direct stimulation of afferents, in vivo. Stimulating afferents with pairs of pulses at a rate of 20 pairs/sec ["tetanus (20 Hz)"] induced PSP depression. This PSP depression was similar to that observed for electrosensory stimuli of the same temporal frequency. Analysis of PSPs elicited by a pair of pulses that preceded versus followed the tetanus revealed that PSP depression was caused by synaptic depression, not by a loss of facilitation. Behavioral studies indicate that fish can detect slow changes in signal amplitude (slow AM) in backgrounds of fast fluctuations. Correspondingly, midbrain neurons respond well to slow AM even in the presence of fast AM. In many neurons, facilitation enhanced responses to trains (8-10 pulses; 100 Hz) that represented activity patterns elicited by slow AM, despite induction of synaptic depression by a tetanus (20 Hz). The interplay between synaptic depression and facilitation, therefore, can act as a filter of temporal information. Some neurons that showed little facilitation nonetheless responded to low temporal-frequency information after induction of depression by fast information; this likely results from the convergence of inputs with different temporal filtering properties.

Integration and recovery processes contribute to the temporal selectivity of neurons in the midbrain of the northern leopard frog, Rana pipiens.

This study examined the mechanisms underlying amplitude modulation selectivity in the anuran auditory midbrain. Single units were recorded extracellularly in the torus semicircularis of the northern leopard frog, Rana pipiens. Two physiologically distinct classes of neurons were identified, based on their response latencies and their selectivities to pulse repetition rates. Cells in one group had short response latencies (median = 31 ms) and responded best to pulse repetition rates below 40 Hz. Tuning to low amplitude modulation rates was largely determined by recovery processes and phasic response properties. Cells in the second group had much longer latencies (median=81 ms) and were generally selective for pulse repetition rates greater than 40-50 Hz. Sensitivity to higher amplitude modulation rates resulted from integration processes; these units only responded when a threshold number of pulses were presented at a minimum pulse density (amplitude modulation rate). At amplitude modulation rates above their best rate, their responses decreased, apparently due to inadequate recovery time between pulses.

Frequency-dependent PSP depression contributes to low-pass temporal filtering in Eigenmannia.

This study examined the contribution of frequency-dependent short-term depression of PSP amplitude to low-pass temporal filtering in the weakly electric fish Eigenmannia. Behavioral and neurophysiological methods were used. Decelerations of the electric organ discharge frequency were measured in response to continuous and discontinuous electrosensory stimuli. Decelerations were strongest (median = 4.7 Hz; range, 3.5-5.9 Hz) at continuous beat rates of approximately 5 Hz and weakest (median = 0.4 Hz; range, 0.0-0.8 Hz) at beat rates of 30 Hz. Gating 20 or 30 Hz stimuli at a rate of 5 Hz, however, elicited decelerations that were sixfold greater than that of continuous stimuli at these beat rates (median = 2.6 Hz; range, 2.0-4.7 Hz for 30 Hz). These results support the hypothesis that short-term processes enhance low-pass filtering by reducing responses to fast beat rates. This hypothesis was tested by recording intracellularly the responses of 33 midbrain neurons to continuous and discontinuous stimuli. Results indicate that short-term depression of PSP amplitude primarily accounts for the steady-state low-pass filtering of these neurons beyond that contributed by their passive and active membrane properties. Previous results demonstrate that passive properties can contribute up to 7 dB of low-pass filtering; PSP depression can add up to an additional 12.5 dB (median = 4.5). PSP depression increased in magnitude with stimulus frequency and showed a prominent short-term component (t(1) = 66 msec at 30 Hz). Initial PSP amplitude recovered fully after a gap of 150 msec for most neurons. Remarkably, recovery of PSP amplitude could be produced by inserting a brief low-temporal frequency component in the stimulus.

Mechanisms for generating temporal filters in the electrosensory system.

Temporal patterns of sensory information are important cues in behaviors ranging from spatial analyses to communication. Neural representations of the temporal structure of sensory signals include fluctuations in the discharge rate of neurons over time (peripheral nervous system) and the differential level of activity in neurons tuned to particular temporal features (temporal filters in the central nervous system). This paper presents our current understanding of the mechanisms responsible for the transformations between these representations in electric fish of the genus Eigenmannia. The roles of passive and active membrane properties of neurons, and frequency-dependent gain-control mechanisms are discussed.

Resolving competing theories for control of the jamming avoidance response: the role of amplitude modulations in electric organ discharge decelerations.

The algorithm for the control of the jamming avoidance response (JAR) of Eigenmannia has been the subject of debate for over two decades. Two competing theories have been proposed to explain how fish determine the correct direction to shift their pacemaker frequency during jamming. One theory emphasizes the role of time-asymmetric beat envelopes, while the other emphasizes the role of amplitude- and phase-difference computations that arise from the differences in spatial geometry of the electric fields of neighboring fish. In repeating earlier experiments, we found that the decision to raise or lower the pacemaker frequency reliably above or below its resting level depends on the latter process, and that frequency deceleration responses to amplitude modulation appear to be sufficient to explain previous experimental results on which the former theory is based. Specifically, fish of the genus Eigenmannia show differential deceleration responses to asymmetric beat envelopes. The deceleration responses do not require phase modulation and show a sensitivity for amplitude modulation depth and selectivity for amplitude modulation rate comparable with that of JARs that are elicited when amplitude- and phase-difference information is available.

Long-term temporal integration in the anuran auditory system.

Analysis of the temporal structure of acoustic signals is important for the communication and survival of a variety of animals including humans. Recognition and discrimination of particular temporal patterns in sounds may involve integration of auditory information presented over hundreds of milliseconds or seconds. Here we show neural evidence for long-term integration in the anuran auditory system. The responses of one class of auditory neurons in the torus semicircularis (auditory midbrain) of frogs reflect the integration of information, gathered over approximately 45-150 ms, from a series of stimulus pulses, not stimulus energy. This integration process is fundamental to the selective responses of these neurons for particular call types.

Passive and active membrane properties contribute to the temporal filtering properties of midbrain neurons in vivo.

This study examined the contributions of passive and active membrane properties to the temporal selectivities of electrosensory neurons in vivo. The intracellular responses to time-varying (2-30 Hz) electrosensory stimulation and current injection of 27 neurons in the midbrain of the weakly electric fish Eigenmannia were recorded. Each neuron was filled with biocytin to reveal its anatomy. Neurons were divided into two biophysically distinct groups based on their frequency-dependent responses to sinusoidal current injection over the range 2-30 Hz. Fourteen neurons showed low-pass filtering, with a maximum decline in the amplitude of voltage responses of >2.6 dB (X = 4.30 dB, s = 1.10 dB) to sinusoidal current injection. These neurons also showed low-pass filtering of electrosensory information but with larger maximum declines in postsynaptic potential amplitude (X = 9.53 dB, s = 3.34 dB; n = 10). These neurons had broad dendritic arbors and relatively spiny dendrites. Five neurons showed all-pass filtering, having maximum decline in the amplitude of voltage responses of <2.0 dB (X = 1.16 dB, s = 0.61 dB). For electrosensory stimuli, however, these neurons showed low-, band-, or high-pass filtering. These neurons had small dendritic arbors and few or no spines. Voltage-dependent "active" conductances were revealed in eight neurons by using several levels of current clamp. In four of these neurons, the duration of the voltage-dependent conductances decreased in concert with the period of the electrosensory stimulus, whereas in the other four neurons the duration of the voltage-dependent conductances was relatively short (<30 msec) and nearly constant across sensory stimulation frequencies. These conductances enhanced the temporal filtering properties of neurons.

Temporal filtering properties of ampullary electrosensory neurons in the torus semicircularis of Eigenmannia: evolutionary and computational implications.

Weakly electric fish have parallel electrosensory systems, the phylogenetically older ampullary system and the novel tuberous system. The tuberous system is an adaptation related to the evolution of active electrolocation. To examine the evolutionary relationship of the ampullary and tuberous systems, the temporal filtering properties of ampullary neurons in the dorsal torus semicircularis of Eigenmannia were studied. 'Whole-cell' recordings were made in vivo using patch-type pipettes. The responses of 19 neurons to sinusoidal electric signals (< 40 Hz) were recorded and the anatomy of these neurons demonstrated by injection of biocytin. All eight low-pass ampullary neurons had broad, relatively smooth post-synaptic potentials (psps) that at low frequencies nicely reflected the sinusoidal stimuli. These neurons had somata of 10-14 microns diameter and thick, spiny dendrites. Eight high-pass neurons were recorded, representing three physiological classes. The first class (3 neurons) had psps that roughly followed the sinusoidal time course of the stimulus; the psp morphology was similar to low-pass neurons. The second class had many small, fast, individual psps; their rate of occurrence varied with the stimulus. Finally, four neurons showed psps that were of constant width across stimulus frequencies. All three classes of high-pass neurons had small somata (8-10 microns diameter) with thin dendrites and either few or no spines. Some of these neurons had large varicosities on the dendrites. Three neurons had band-pass filtering properties: neurons that showed strong band-pass properties were morphologically similar to low-pass neurons. Comparisons of the temporal filtering, shapes of post-synaptic potentials, and anatomy of ampullary and tuberous neurons in the torus suggest that the circuitry for tuberous processing in the torus may have evolved as an elaboration or duplication of the ampullary system. The mechanisms underlying the low-pass filtering characteristics of tuberous neurons therefore appear to have predated the evolution of the tuberous system and to have served as a pre-adaptation for the evolution of the jamming avoidance response. In addition, these data support the hypothesis that spine density influences the temporal filtering properties of neurons.

New techniques for making whole-cell recordings from CNS neurons in vivo.

Patch-type pipettes increasingly are being used to obtain intracellular 'whole-cell' recordings from neurons. Here we describe our methods for making whole-cell recordings in vivo from midbrain neurons in an electric fish. Novel elements in the procedure are: A device for micropositioning the pipette when near a cell, use of a 'Picospritzer' for cleaning the pipette tip and cell surface, and an electroporetic method for perforating the patch following seal formulation. In addition, we show that extracellular and intracellular recordings can be made from the same neuron. Stable intracellular recordings can be made from neurons at least as small as 10 microns.

Hierarchical sensory guidance of mauthner-mediated escape responses in goldfish (Carassius auratus) and cichlids (Haplochromis burtoni).

Acoustically-evoked escape behaviors were compared between goldfish (Carassius auratus), a hearing specialist, and the cichlid Haplochromis burtoni, a hearing nonspecialist. Fish were startled with compressive and rarefying, stimuli presented alone or together, and with compressive pulses preceded by a visual cue or after exposure to cobalt, an inhibitor of lateral line-innervated neuromast hair cells. These acoustic startle stimuli can evoke Mauthner neuron firing and are similar to but weaker than those produced by a largemouth bass (Micropterus salmoides) feeding on guppies. When sound stimuli were presented alone, both species avoided the direction of either the compressive or rarefying stimulus. If a light preceded and was contralateral to the compressive sound pulse, goldfish continued to avoid the sound source; cichlids avoided the visual cue and turned toward the sound. Goldfish performance improved significantly when the visual cue was in the same direction as the sound source. Goldfish performance also improved significantly after exposure to 0.1 mmol l-1 cobalt solution for 24 hours before testing, but cichlids would not startle after cobalt acclimation. A compressive pulse presented to one side of a fish simultaneously with a rarefying pulse on the other side causes the entire fish to accelerate with the water current. This strongly and directly accelerates the ear but tends to reduce both the pressure changes transduced by the swimbladder and activation of the mechanosensory lateral line. In this test, goldfish reliably avoided the compressive pulse. Cichlids, however, randomly avoided either speaker polarity but significantly avoided the speaker which had a faster onset. With more closely matched speakers, cichlids also preferentially avoided the compressive stimulus. Thus, the primitive sensory condition for auditory activation and guidance of Mauthner-neuron-initiated escape responses may have evolved to detect the initially compressive sounds associated with ram-type predators.

Temporal filtering properties of midbrain neurons in an electric fish: implications for the function of dendritic spines.

Electrosensory neurons in the torus semicircularis (midbrain) of the weakly electric fish Eigenmannia vary considerably in their dendritic structure and responses to modulations of the amplitude of electric organ discharges. We investigated possible relations between these properties by recording intracellularly and labeling individual neurons while modulating stimulus amplitude over rates of approximately 2-20 Hz. Morphologically distinct cell types generally differed in their responses to these stimuli. The amplitude envelope of the stimulus was nicely reflected in fluctuations of the membrane potential of heavily spined neurons. The amplitude of these stimulus-related depolarizations decreased markedly as the stimulus modulation rate was increased. For aspiny or sparsely spined neurons, however, the amplitude of stimulus-related depolarizations either increased or remained constant over this range of modulation rates. In these cells, the amplitude envelope of the stimulus was not well represented in the membrane potential. Instead, fast EPSPs were observed that varied in number over time in accordance with the amplitude envelope of the stimulus. Aspiny neurons in the tectum also coded the amplitude envelope of stimuli with poor fidelity. The amplitude of stimulus-related depolarizations, however, decreased as the rate of modulation of stimulus amplitude was increased, consistent with the notion that tectal neurons receive afferent input from the spiny toral neurons. Spiny neurons appear, therefore, to act as low-pass filters of temporal information in sensory signals. Aspiny cells, however, code high temporal frequencies. These data support the hypothesis that dendritic spines contribute to the low-pass filtering of inputs to neurons.

Evidence for the role of dendritic spines in the temporal filtering properties of neurons: the decoding problem and beyond.

The question of relations between structure and function acutely applies to the search for the functional role of dendritic spines. While dendritic spines are a prominent and widespread structural feature of neurons in the central nervous system, their function is poorly understood. Because the conducting core of a spine stem can be of extremely small dimensions, a large axial resistance to current flow and "low-pass" filtering of inputs have been hypothesized. Here we show that neurons in the dorsal torus semicircularis of the electric fish Eigenmannia show real-time fluctuations in their transmembrane potential that reflect modulations in the amplitude of a high-frequency sinusoidal carrier signal. In 18 neurons recorded intracellularly and labeled with Lucifer yellow, the decrease in the magnitude of these potentials with increasing rate of amplitude modulation (i.e., low-pass temporal filtering) was positively correlated (r = 0.79, P < 0.001, over a range of one to two octaves in modulation rate) with the mean dendritic spine density (range, 0-0.38 spine per micron of dendritic length) of the cell. The acquisition of synaptic input through dendritic spines may be a general mechanism for achieving the temporal filtering that underlies real-time signal processing in the central nervous system.

Differential distribution of ampullary and tuberous processing in the torus semicircularis of Eigenmannia.

Gymnotiform electric fish sense low- and high-frequency electric signals with ampullary and tuberous electroreceptors, respectively. We employed intracellular recording and labeling methods to investigate ampullary and tuberous information processing in laminae 1-5 of the dorsal torus semicircularis of Eigenmannia. Ampullary afferents arborized extensively in laminae 1-3 and, in some cases, lamina 7. Unlike tuberous afferents to the torus, ampullary afferents had numerous varicosities along their finest-diameter branches. Neurons that were primarily ampullary were found in lamina 3. Neurons primarily excited by tuberous stimuli were found in lamina 5 and, more rarely, in lamina 4. Cells that had dendrites in lamina 1-3 and 5 could be recruited by both ampullary and tuberous stimuli. These bimodal cells were found in lamina 4. During courtship, Eigenmannia produces interruptions of its electric organ discharges. These interruptions stimulate ampullary and tuberous receptors. The integration of ampullary and tuberous information may be important in the processing of these communication signals.

Abstract art and emotion: expressive form and the sense of wholeness.

This paper limits itself to abstract art the better to concentrate on the relation between the emotionally expressive power of esthetic form--apart from its narrative content--and emotional responsiveness. The emotionally expressive power of art--not to be confused with the artist's own emotions--has to do with the way sensuous esthetic forms highlight the rhythmic changes of tension and release inherent in ordinary perceptual experience. Tension and release are useful terms in thinking about how the perception of an esthetic structure is transformed (transduced) into feelingful psychological meanings, and contributes to the subjective sense of wholeness. The sense of wholeness may be illusory and/or authentic depending on the mix of elements in the individual's responsiveness. It comes about through a unified organization of tension and release, as embodied in expressive forms such as rhythm and rhyme, resonating with tension and release evoked in the observer's associations to such psychological issues as separation and reunion. Having stirred the viewer's emotional responsiveness, the art work provides a reliable "container" for the objectification of latent emotions.

Anatomical and functional organization of the prepacemaker nucleus in gymnotiform electric fish: the accommodation of two behaviors in one nucleus.

The diencephalic prepacemaker nucleus (PPn) of gymnotiform electric fish projects to the medullary pacemaker nucleus and modulates its regular firing frequency. Each firing of the pacemaker, in turn, drives an electric organ discharge (EOD). Two types of PPn neurons were retrogradely labeled from the pacemaker with HRP in Eigenmannia and Apteronotus. In both species, smaller ovoidal cells were found in the dorsomedial part of the PPn (PPn-G), and larger multipolar cells were found in the ventrolateral part of the PPn (PPn-C). This morphological distinction between the two subnuclei in the PPn was paralleled by a functional dichotomy. Microiontophoresis of L-glutamate in the PPn-G of both species elicited slow and gradual accelerations of EOD frequency characterized by a time constant on the order of seconds. The elicited frequency modulations were similar to those observed during the jamming avoidance response and during courtship. Glutamate stimulation of the PPn-C, in contrast, produced fast and abrupt frequency modulations characterized by a time constant on the order of milliseconds. These abrupt modulations resembled "chirps" observed during courtship and aggression. Similar behavior was produced by intracellular current injection into a PPn-C neuron of Apteronotus, and intracellular labeling of this neuron with Lucifer Yellow revealed a multipolar PPn-C neuron similar to those retrogradely labeled with HRP.

'Recognition units' at the top of a neuronal hierarchy? Prepacemaker neurons in Eigenmannia code the sign of frequency differences unambiguously.

The electric fish, Eigenmannia, is able to discriminate the sign of the frequency difference, Df, between a neighbor's electric organ discharges (EODs) and its own. The fish lowers its EOD frequency for positive Dfs and raises its frequency for negative Dfs to minimize jamming of its electrolocation ability by a neighbor's EODs of similar frequency. This jamming avoidance response (JAR) is controlled by a group of 'sign-selective' neurons in the prepacemaker nucleus (PPN) that is located at the boundary of the midbrain and the diencephalon (Fig. 1). Extracellular recordings from a total of 35 neurons revealed a great similarity between behavioral and neuronal response properties: 1. All neurons fired vigorously for negative Dfs and were almost silent for positive Dfs, regardless of the orientation of the jamming stimulus, and thus discriminated the sign of Df unambiguously (Fig. 2). 2. In accordance with behavioral observations, individual neurons failed to discriminate the sign of Df when the jamming stimulus had the same field geometry as the signal mimicking the animal's own EOD (Fig. 3). 3. Df magnitudes which evoke strongest JARs, usually 4 to 8 Hz, also induced most vigorous responses in sign-selective neurons (Fig. 5). 4. Behavioral and neuronal thresholds for the detection of small jamming signals were similar. Threshold for sign selectivity was reached when the amplitude ratio of the jamming signal to the EOD mimic, measured near the head surface, was 0.001. This value corresponds to a maximal temporal disparity (a necessary cue for performing a correct JAR) of 1 to 2 microseconds for signals received by the two sides of the body in a transverse jamming field (Fig. 7). 5. The effects of two jamming fields, offered orthogonally to each other, may interact nonlinearly at the behavioral as well as at the neuronal level. A positive Df presented in one field may suppress behavioral and neuronal responses to modulations of the sign of Df in the other field (Fig. 8c).