Neuronal Coding of Perceptual Systems

The following sections are included:

• PREFACE

• CONTENTS

One of the biggest challenges in understanding perception is to understand how the nervous system manages to integrate the multiple codes it uses to represent features in multiple sensory modalities. From different cortical areas, which might separately register the sight of something red and the touch of something smooth, one effortlessly generates the perception of one thing that is both red and smooth. This process has been variously called “feature integration”, “binding”, or “synthesis”. Citing some current models and some historical precursors, this paper makes some simple observations about the logic of feature integration. I suggest that “feature conjunction” is not strictly speaking conjunction at all, but rather joint predication; and that the critical task in “binding” is not simply grouping scattered representations together, or providing them a common label, but rather identifying those that have a common subject matter - those that are about the same thing. If this is correct, it follows that the vocabulary of sense includes not only features but something analogous to referring terms.

In this article, 1) the method of psychophysiological simulations is described and discussed. 2) The physiological steady-state models of photoreceptors and colour opponent coding (COC) including light adaptation and of the colour choice behaviour of the bee are reviewed in short. 3) Simulations of co-evolution of neuronal colour coding systems of pollinating insects and optical properties of flowers are reviewed and discussed. As is actually the case in bees and other insects, brightness coding neurones did not develop in the simulations, because a colour brightness dimension turned out to be a disadvantage for nectar finding and pollinating flowers. Polychromatic colour vision systems tended to degenerate to oligochromatic systems. The flower colours developed for best visibility and discriminability for the colour vision systems. 4) In addition, extensions of the photoreceptor model and the COC model of the bee, describing the fluctuations in the membrane potentials by Monte-Carlo simulation, are reviewed and discussed. The slight light-intensity dependence of the fluctuation ranges in the photoreceptors are levelled out by colour opponent coding. The fluctuations in the decision making neurones, on which the colour choice behaviour relies, turned out to be intensity independent, as measured in behavioural experiments (Thurstone’s Case V). 5) Furthermore, the photoreceptor and COC models of the bee were extended by temporal properties, including the light adaptation processes. The obtained physiological adequate models of the photoreceptors, monopolar cells, and colour opponent coding (COC) neurones including their synapses are described in review and applications are discussed. The temporal colour space, spanned by the COC neurones, allows us now to describe and understand which colours bees and other insects actually perceive in their environments, over time, during slow and fast flight.

Normal humans can match any wavelength of light with a combination of three other wavelengths. This capacity, called trichromacy, is due to the presence of three types of cone photopigment in the photoreceptors. These photopigments differ in their peak absorption at either short (S), middle (M), or long (L) wavelengths. A match between two fields of differing spectral distribution occurs when the relative intensities of lights in one field are adjusted so that the activity in each of the three classes of photoreceptor is identical for the two fields. This is possible because of the principle of univariance: light absorbed by a photopigment molecule produces the same response independent of wavelength.

Bipolar cells separate the response of the photoreceptors, giving rise to parallel ON and OFF pathways from the retinal ganglion cells to cortex. The main projection site of the retinal ganglion cells for processing of chromatic signals is to the lateral geniculate nucleus (LGN). One class of retinal ganglion and LGN cell (parvocell) carries signals based on combinations of cone signals, L-M or S-(L+M). Although neural signals carrying colour information undergo additional transformations in the cortex, responses of these cells can account for many aspects of colour discrimination.

Large color differences between lights at two photopic levels were studied in colornormal (NT) and color-deficient (CVD) observers. The MDS configuration in a common color space was a hypercylinder, with the 4^th, Saturation, dimension allowing the ends to ‘bulge out’ into spherical surfaces. This dimension was dependent on the chromatic-axes values. In contrast to the NTs, for the CVDs ‘Brightness’ dimension was more salient than Saturation. Spaces for the CVDs were flattened hypercylinders, with elliptical ends. Within the CVD group, slightly different axes of compression distinguished protans and deutans.

Measurements of spectral sensitivity, wavelength discrimination and additive color mixture using a two-choice training technique showed that the goldfish has a highly effective tetrachromatic color vision system which is based on four cone types. The spectral sensitivity function obtained by training the goldfish on the dark test field with the second test field illuminated with monochromatic light showed pronounced maxima and minima. Shape and location of the maxima were very similar to those found in recordings from horizontal cells and ganglion cells in goldfish and carp retina. They indicate inhibitory interactions between cone channels of different spectral type. A very different spectral sensitivity function was found when the goldfish were trained on the illuminated test field while the comparison test field was dark. In this case, in which a broad function was obtained indicating the sum or the over-envelop of cone sensitivity, goldfish were probably not discriminating on the basis of “color”, but of “brightness”. Experiments showed that with the drug ethambutol and under mesopic light conditions the L-cone contribution to color vision disappeared. The goldfish became red-green color blind. This was also the case when a dopamine-D l antagonist was injected into the eye. However, Lcone contribution was not affected in the context of brightness and motion vision. These and other results indicate that there are separate and parallel channels for processing of “color” on the one hand, and “brightness” and “motion” on the other hand.

Insect eyes combine several mechanisms to achieve and improve color vision. These include the expression of various visual pigments having different absorption spectra, packing the visual pigments in optical waveguides, and applying different optical filters consisting of either spectral selective pigments or interference filters and reflectors. This chapter presents a few examples of optical filters in the eyes of butterflies and horseflies. Optical multilayers exist in the ommatidia of butterflies, proximal to the photoreceptor cells, acting as spectrally selective reflectors, presumably for enhancement of light sensitivity and color discrimination. In horseflies multilayers in the corneal facet lenses act as selective transmission filters, probably resulting in modulation of the sensitivity spectra of the underlying photoreceptors.

A physiological model of the dark and light adapted photoreceptors of the honeybee worker (Apis mellifera) has been developed. Also the parameters of the model were fitted exclusively to measured membrane potentials of a dark adapted photoreceptor, the model also accurately describes light adapted photoreceptor membrane potentials. Furthermore, the model correctly, predicts the potential time courses of measured photoreceptor responses with respect to square-wave modulated flicker lights of up to 150 Hz frequency. The performed tests clearly demonstrate that the presented photoreceptor model indeed is a physiologically adequate description of essential components of the phototransduction and the electrical membrane processes in the photoreceptors of the honeybee worker.

The optical system of the eye produces a 2-dimensional image that is sampled by a huge array of photoreceptors. In capturing, processing, and encoding the spatial information at the inital levels (within the eye), the visual system must solve two major problems and operate with three serious constraints. Many aspects of our spatial vision reflect the way in which the visual system goes about solving these problems with these constraints.

One major problem is that of transducing the tiny energy of single photons into neural activity which involves the relatively massive flow of ions across cell membranes. A second problem is that of separating the tiny amount of information in the image about reflective objects (the useful information) from the massive variations in light due to the illuminant (relatively useless information), when the two are totally confounded at each point in the image. One serious early constraint on spatial vision is the bottleneck of the optic nerve, which severely restricts how much information can be sent from the eye to the brain. A second constraint is that the system must simultaneously process and encode color and spatial information, with different requirements for the two and the need to send both kinds of information up through the optic nerve bottleneck. A third constraint is the very limited capacity of axons to carry information through spike firing rates while operating in real time.

The receptors amplify the varying photon-capture signal with an enzymatic cascade that provides temporal integration at the cost of a slow response and low temporal resolution. To start the separation of illuminant from reflective object, the visual system attenuates very low temporal and spatial frequencies, in the retinal processing. It requires many more neurons to encode high than low spatial frequency information, so the visual system processes and transmits high spatial frequency information only over a small retinal region. Furthermore, spatial information is largely multiplexed with color information to maximize information transmission through the optic nerve bottleneck, and there are parallel systems encoding increments and decrements (and color changes in opposite directions) around the mean illumination level.

Stimulation of the vertebrate retina by a step of light elicits a prolonged excitatory post-synaptic current (EPSC) in sustained amacrine cells and continuous spiking during the step. In transient amacrine cells, the same step stimulus instead elicits a very transient EPSC and a rapid burst of spikes at stimulus onset and offset. Biophysical and morphological studies have revealed many of the important data necessary to make a formalized compartmental model of the connections and events leading to the conversion of a sustained light-step stimulus into a transient EPSC and spike train in transient amacrine cells. Using a compartmental neural network model that incorporates data from patch clamp and imaging studies in the retinal slice preparation, we are implementing a model of the amacrine cells and their synaptic connections within the inner plexiform layer that simulates the response features of both transient and sustained amacrine cells. The model is implemented using a new compartmental modeling system called CONICAL. The model describes fundamental features in the sustained and transient amacrine cells that are necessary for their respective response patterns.

Simple cells in the primary visual cortex respond selectively to oriented stimuli. It has been proposed that such feature detecting neurons should generate a sparse representation of the visual world and orientation selective receptive fields are in this sense optimal spatial filters for “natural” visual environments. In this contribution we show that a competitive Hebbian development model driven by natural images may generate a topographic projection from the lateral geniculate nucleus to the primary visual cortex as well as orientation selective receptive fields. The resulting representation is sparse and the degree of sparseness depends on the recurrent dynamics, i.e., the lateral competition among the cortical neurons. Simulations show that for weak competition, the resulting receptive fields are global and unstructured and for intermediate competition, they refine and a topographic projection emerges. Finally, in the case of strong competition, the receptive fields act as oriented localized spatial filters arranged in a typical orientation selectivity map.

Adult male semiterrestrial crabs were trained monocularly and then tested after 24 hours, in order to study interocular transfer (IOT) in two associative memory paradigms: 1) Waning of the escape response (ER) to an iterated visual danger stimulus and 2) Waning of the exploratory activity to a novel environment. In both paradigms monocular crabs exhibited retention after 24 hours when trained and tested with the same uncovered eye, but not with different uncovered eyes. Since acquisition is achieved by monocular crabs, this failure of IOT may be explained in terms of retrieval impairment due to a) a different perception of the training and testing stimuli, an hypothesis that does not rule out IOT, b) lateralization of memory storage or c) the existence of two symmetrical, independent and redundant lateral storage sites.

The data on physiology of the visual sensory system and on neurophysiological mechanisms of the rhythmic EEG activity suggest that transformation of the pattern of unit activity have to disrupt the information structure of afferent and central spike volleys with the phasic high frequency spike bursts and inhibitory pauses, and by prolonged tonic inhibition of principal cell’s discharges due to the hyperpolarizing shift of their membrane potential, as well as by synchronous recruitment into rhythmic activity of neighbouring neurons - “detectors” of different sensory stimuli features. Tlie set of experimental data on relations between human EEG alpha and impairments of visual perception, and of quality of visuomotor pursuing performance is presented supporting the hypothesis that adequate information processing in sensory systems have to be significantly impaired or suppressed during rhythmic EEG activity.

The middle temporal (MT) visual (V5) area is, together with its projection targets, one of the main regions involved in processing of retinal motion in the monkey. Its homologue in humans has been identified. The receptive field (RF) often contains suppressive parts referred to as the antagonistic surround. Many MT/V5 neurones are selective for direction and speed of translation. They are also selective for direction of speed gradients corresponding to tilt direction, but not for kinetic boundaries nor optic flow components. These properties have used to predict responses in human activation studies.

Desert ants like honey bees can perceive the pattern of polarized light (E-vector pattern) in the sky and use it for navigation. This truly exceptional task accomplished by the insect’s nervous system, has challenged the ingenuity of neuroethologists for more than two decades. It was assumed that the insect solved the problem in an allinclusive way, i.e. that it was capable of (i) determining locally the orientation of individual E-vectors in the sky, and (ii) “knowing” where these individual E-vectors occurred within a sky-based system of reference. In contrast, our combined behavioural and neurophysiological work has shown that the insect uses a small dedicated system to solve the problem in an approximate yet completely sufficient way. Small-field Evector detectors (micro-analyzers) project on to a few large-field interneurons (macroanalyzers). The points of the skylight compass are encoded in the response ratios of three types of macro-analyzers. Potential ambiguities and imperfections are eliminated by wide-field integration and time-and-again recalibration of the system. The hypothesis derived from our neuroethological analyses is tested by implementing it into an autonomous agent, a robot, which navigates along-side the Cataglyphis ants in the North African desert habitat.

Although it is known that the goldfish can discriminate color in the ultraviolet range, to date there is no electrophysiological data to reveal the neural basis of this discrimination. In the present study, intracellular responses to monochromatic stimuli including the UV range were recorded from the outer retina of the goldfish. Recordings were performed on isolated retinas, using borosilicate microelectrodes filled with 3M KCl (100-500MΩ). Each cell was tested with monochromatic stimuli of equal quanta (300-700 nm). The size of the receptive field, as well as the response/intensity function were also assessed in different spectral regions, in order to calculate the spectral sensitivity of the studied cells. A triphasic bipolar cell type with depolarizing responses from 300 to 400 nm and 520 to 700 urn (UV+G+R+/B−) was found. These cells (n=2) also show spacial opponency in their receptive fields, with a typical center/surround organization. This pattern corresponds, from 400 nm on, to that of the already known biphasic colour-coded bipolar cells. The opponency between the ultraviolet and violet spectral regions, however, shows that this cell type is in fact triphasic. This might be the first step in the neural mechanism for discrimination of ultraviolet light.

To study processing of UV stimuli in the inner retina of the turtle, we recorded intracellular responses to spectral light from 33 amacrine and 34 ganglion cells. All amacrine and ganglion cells responded to UV stimuli. One color opponent amacrine cell hyperpolarized to UV and violet spots, but depolarized to green and red light. One off ganglion cell was maximally inhibited by UV light spots and was excited by annuli, regardless of wavelenth. A second ganglion cell depolarized to UV light and hyperpolarized to green and red light. The results indicate a specific UV input to retinal neurons in the inner retina of the turtle.

A sense of hearing has evolved more than twenty times within the vertebrates and insects, making use of a wide variety of physical mechanisms. While most ears respond to the pressure component of sound, many hearing organs detect the oscillatory flows of the medium, which is most prominent close to the sound sources. It is argued that it would be absurd to use the term hearing only for responses to the pressure component. Most pressure-sensitive ears receive sound by means of an ear drum, the vibrations of which may either be guided to an inner ear via middle ear ossicles (terrestrial vertebrates) or detected by receptor cells attaching to the ear drum (e.g., moths, grasshoppers). However, in some cases there is no obvious connection between the ear drum and the inner ear. While some animals like the insect prey of echo-locating bats do not need to analyse sound frequency, most animals using sounds for social communication have a capacity for frequency analysis. Most hearing animals are able to determine the direction to the sound source, but the physical mechanisms used for directional hearing are very diverse.

Transduction of mechanical energy to electrical membrane responses has the characteristics of direct mechanical control of ion channels, at least in the sensory cells studied, as is appropriate for high time resolution of stimuli; while in some other mechanosensitive cells channels may be controlled more indirectly. In epithelial mechanoreceptor cells, present in all metazoa, the stimulus receiving cell organelles are cilia and/or microvilli (stereovilli) at the accessible apical side of the cell, in four different configurations: monociliary cells with a concentric or eccentric bundle of stereovilli (in most invertebrates or vertebrates, resp.; reduced cilium in the organ of Corti), multiciliary cells with microvilli (in molluscs), and monociliary cells without microvilli (in arthropods). In any case, force is concentrated by extracellular connectors to small areas of ciliary and/or micro-(stereo-)villar membrane. In some cases opposed intracellular bridges (membrane-integrated cones) were specified that connect this area to the cytoskeleton. Blockage of transduction by antibodies directed against the connectors suggests that mechanosensitive ion channels are controlled by the trans-membrane chains composed of the connectors and bridges. The cellular and sub-cellular organisation funnels the stimulatory energy to the few trans-membrane molecules. –Each of the four basic cell configurations have been varied independently in evolution to accommodate to the various sources of mechanical energy, leading to some convergent organisations of organs.

Ensemble coding of differently oriented straight lines and geometrical shapes were studied in man. The movements were imposed at the level of the ankle. Muscle spindle afferents originating in four ankle muscles were recorded from the lateral peroneal nerve using the microneurographic technique. The data were analyzed in two different ways: a qualitative approach (describing how the activity of these afferents covary in time and space), and a quantitative approach, (i.e., a “vectorial model”). This vectorial model is based on the idea that such neuronal coding can be analyzed in terms of a series of vectors expressing some movement parameters. In our case, each vector represents the mean contribution of all the muscle spindle afferents within one muscle. A particular vector points in the “preferred sensory direction” of the muscle it represents, and has a length corresponding to the mean frequency of all the afferents within that muscle. The hypothesis is that the sum of these weighted muscle afferent vectors points in the same direction as the ongoing movement. This paper will make the point that sufficient proprioceptive information about movement trajectories can only be attained if the whole “sensory landscape” (i.e., the sensory activities of all the muscles involved in the movement) is taken into account to subserve kinaesthetic perception.

In order to study the organization of the proprioceptive sensory codes subserving movement perception, muscle spindle activities were recorded in humans using the microneurography method during actual movements. On this basis, complex handdrawing illusions were elicited using various vibration patterns applied to the wrist muscles of human subjects. It was established that it is possible to elicit kinesthetic illusions involving spatially oriented lines and geometrical shapes such as rectilinear or curvilinear figures by activating four groups of muscle tendons at the wrist level. The vibration sequences specifically evoking each shape were determined by varying the vibration frequency, the duration of each stimulus and the vibrator onsets, and by applying the vibrators either successively or simultaneously. The proprioceptive coding of a trajectory can be modeled in terms of a series of vectors, whose direction depends on the anatomical sites of the muscles stretched and shortened during the movement. The vector giving the spatial path of a movement is the sum of the vectors of the proprioceptive inputs originating from each muscle, and the modulus of the resulting vector is the instantaneous velocity of the movement. It therefore emerges from the results of this study that muscle proprioception is able to generate spatio temporal afferent patterns that may mediate complex cognitive operations such as those involved in the memorizing and recognition of motor forms. Moreover, some experimental data indicate that proprioceptive inputs from all the muscles holding and moving the retina in space, from the foot up to the eyes, are used by the brain to process the visual information required to perform spatial localization and reaching tasks. It is suggested that proprioception may provide a link between personal and extrapersonal space.

This paper illustrates results from an acute cat model showing the effects of muscular fatigue on the neuronal coding capacity in ensembles of primary muscle spindle afferents. Furthermore, the results are related to the level of consciousness by investigating the effect of muscular fatigue on proprioceptive ability in human subjects.

In the cat model, the neuronal coding capacity was evaluated using the method for analysis of encoding of stimulus separation described by Johansson and co-workers. The results from this study show that fatigue of the ipsilateral medial gastrocnemius muscle caused a clear-cut reduction in the ability of ensembles of primary muscle spindle afferents from the lateral gastrocnemius muscle to discriminate between muscle stretches of varying amplitude.

In the studies on human subjects, alterations in the movement sense acuity during localized muscle fatigue in the dominant shoulder was investigated. This study demonstrated a lower probability of subjects distinguishing between different movement velocities following hard exercise as compared to light exercise conditions (P<0.001).

It is well known that primary muscle spindle afferents play an important role in proprioception and kinaesthesia. Therefore the decrease in the accuracy of the information transmitted by ensembles of primary muscle spindle afferents, as seen in the cat model, might very well explain the decrease in proprioceptive acuity observed in the human studies.

A new method was used for analysis of ensemble coding in populations of receptor afferents. The aim was to investigate ensemble coding in populations of primary muscle spindle afferents (MSAs) and to assess the role of the fusimotor system for the encoding ability of these ensembles. Secondly, to investigate ensemble coding in mixed populations of different muscle afferents, and to compare the capacity of mixed populations with populations of only one type of afferent. Thirdly we studied the effects of muscle fatigue on the stimulus separation in ensembles of primary MSAs. The results show that ensembles of primary MSAs discriminated better between muscle stretches than individual MSAs. Mixed ensembles of afferents discriminated better between muscle stretches than ensembles of only one type of afferent. The increase in separation with increasing ensemble size was considerably reduced after cutting the ventral roots or induction of muscle fatigue in heteronymous muscles, signifying the importance of an intact fusimotor drive for stimulus separation in ensembles of primary MSAs.

It is not easy to illustrate 2D drawing figures as variations in time of X and Y co-ordinates in normal static figures. For that reason, we developed a computer controlled audio-visual software permitting the audience to follow, in real time, the ongoing drawing movement together with two individual muscle spindle spike trains (e.g., Efrom an ago/antagonist pair of muscles). Furthermore, each spike was converted to an auditory signal similar to that heard during actual neurophysiological experiments. With this software, it was possible to get a global understanding of how a complex drawing movement (such as square, circle, triangle etc…) is coded by the sensory system in humans.

Sensory coding taking place at the level of receptor cells, and the adjacent neuronal circuitry is a conditio sine qua non for processing of olfactory (and taste) stimuli. However, those sensory events do not represent the whole coding process. Different features concerning the stimuli in question are being “ascribed” to the corresponding neural message processed by the brain, probably. The added parts of the code are derived from two sources, apparently: from information concerning the proper stimulus (feature coding), and those generated from interactions between the stimulus and the brain (interaction coding). The results discussed in the present paper have been acquired (with a single exception) in psychophysical experiments. The aim was to draw inferences on the underlying coding processes. Candidates for feature codes could be considered (i.a.): the presence/absence of the stimulus, its quality, intensity, spatial localization (laterality), discriminability from other stimuli, mutual timing with respect to other stimuli, adherence to a stimulus sequence of increasing intensity. Interaction coding concerns (i.a.): the conscious nature of the stimulus, its hedonic interpretation, the signalling features added by Pavlovian (associative) conditioning, and its dependence of the actual state of the brain.

In perpheral taste, the coding mechanism remains an enigma. There are two theories: the across-fiber pattern theory argues “that a particular pattern of activity across the entire ensemble of afferent fibers represents a taste quality”, while the labeled-line suggests “that activity in a particular fiber type represents a specific taste quality”. This study summarizes the results from taste nerve recordings and behavior in a number of mammalian species. It links the sweet taste quality to activity in a fixed set of taste fibers forming an S-cluster. Thus our findings seem to satisfy the definition of the labeled-line theory: “that activity in a particular fiber type represents a specific taste quality“ (Smith & Frank, 1993). However, it is possible that within the S-cluster taste information is being extracted across the fibers by the CNS.

Insects detect small changes in odour concentration at astoundingly low thresholds and recognize and distinguish complex blends of odours. Furthermore, they are able to localize odour sources within short range as well as over long distances. Specialized olfactory receptor neurons on their antenna and neuronal processing in the brain of insects allows for the detection of the biologically relevant properties of chemical stimuli of the environment and for the selection of an appropriate behavioral response. Here, a brief summary is given of the structure and function of the peripheral olfactory system of insects.

Temperature is a dominant factor in the function of biological systems. Most animal species therefore have thermoreceptors that detect changes in the temperature of their body parts. Whereas in vertebrates primary sensory neurons acting as temperature sensors are equipped with heat-sensitive ion channels, the thennoreceptors of insects probably are modified mechanoreceptive cells. Heat-induced mechanical changes probably also occur in the infrared receptors of fire-loving beetles. Their specialized pit organs detect infrared radiation, emitted by objects with a high temperature. Thennoreceptors are multimodal as they respond also to chemical or other physical stimuli. Although the detection mechanism of visual photoreceptors fundamentally differs from that of the thermoreceptors, photoreceptors are strongly temperature dependent, resulting in a higher temporal acuity with increasing temperature. Distinct changes in body temperature can occur in flying insects, that are measurable with a thermal imaging camera. In large, flying insects this requires active thermoregulation, where the thermoreceptors presumably play an important role.

Eigenmannia virescens is a South American freshwater species with an electric organ discharge (EOD) of the wave type. This social fish’s EOD is masked by the intense noise from close-by conspecifics. In spite of this, Eigenmannia detects external signals at lowest thresholds for a vertebrate octavolateralis sensory system; furthermore, Eigenmannia discriminates between stimulus waveforms which is unknown for mechanical sensory modalities. It is suggested that an analysis of beat signals in three steps underlies these sensory feats: (1) stimulus filtering and intensity assessment of a single harmonic of an external signal, in spite of only weak tuning of the highfrequency part of the electrosensory system. (2) Assessment of the frequency difference between a fish’s own EOD and that of an external signal. It is shown that temporal cues of the mixed (beat) signal are sufficient, and that beat amplitude cues are not required; nor do they seem to be used by the fish. Steps 1 and 2 completed, a fish may change its EOD frequency (e.g., by a jamming avoidance response) for optimizing (3) the assessment of stimulus waveform from the beat signal, such as external female and male EODs. Sensory models for the assessment of frequency difference and stimulus waveform are presented.

The use of the geomagnetic field in orientation and navigation was first reported in the 1960s. Today, behavioural evidence indicates that magnetic orientation is rather widespread among animals of various systematic groups. Two different types of magnetic compass mechanisms have been described, and the use of magnetic parameters as components of the navigation ‘map’ is considered. However, neither the primary processes of magnetoreception nor the location of receptors and the processing centres are more than vaguely known. One of the hypotheses currently discussed proposes light-dependent processes of magnetoreception; they involve excited state macromolecules and are assumed to take place in the retina. The other hypothesis assumes magnetoreception mediated by magnetic particles as they are found in the ethmoid region of many vertebrates. Both hypotheses are supported by first behavioural findings. In electrophysiological studies, responses to changes in the direction of the ambient field were recorded in parts of the accessory optic system and the tectum opticum of birds; they are discussed as providing directional information for the magnetic compass. Electophysiological responses to changes in intensity were reported from the ophthalmic branch of the nervus trigeminus in birds and fish and from the trigeminal ganglion of birds. In birds, this input is discussed as providing information for the position-finding system of the navigational ‘map’.

Pain is a complex biological phenomena that is mediated by a number of interacting systems. Nociceptive responses to various aversive stimuli have been used to examine pain mechanisms and sensitivity. Nociceptive responses to noxious stimuli are not invariant, with there being both pain inhibitory and facilitatory mechanisms. Prominent among these pain inhibitory or analgesic systems are the endogenous opioid peptides and their associated regulatory mechanisms. Included among the factors that can influence opioid modulation of nociceptive sensitivity are magnetic stimuli. The present review discusses current evidence relating pain sensitivity, opioid activity and the effects of magnetic fields.

A neural oscillator capable of processing graded inputs is studied. The oscillator has two functional modes controlled by an external signal and codes information either by the amplitude of its oscillations or by the coordinates of its fixed point. Excitatory and inhibitory connections between coupled oscillators control their phase relations. Simulations and theoretical analyses show that any desired phase relation can be induced by an appropriate choice of connections.

Throughout the nervous system we find ‘projective’ fields, ‘projective’ tracts and mirrored ‘projections’. We shall try to calculate such projections using nerve pulses in physical dimensions. By careful control of the parameters, mirrored pulse interference projections can be simulated. We find overlay, interference overflow, and moving and zooming effects. Neural data addressing appears in short-circuit connectivity. Introducing new, physical basic functions of a neuron, bursts can be identified as neural addresses. A physical interpretation of pain is introduced. The paper offers a wave-theoretical approach for calculating the behaviour of short-circuit networks, trying to avoid reference to synaptic weights.

This paper investigates the use of Weightless Neural Networks (WNNs) for pattern recognition. In particular, this work focuses on grey level image processing issues related to the WNN.The coding of the grey level image into a binary one is a crucial step for the WNN architecture. For the recognition process, it is important that similar grey images should correspond to similar binary patterns. For this reason, various approaches have been used to compare different processing methods and an original part of this work is the use of new quantitative approach to analyse similarities between images and between their binary representations. In this paper the advantage of using Minchinton Cells, one of the coding methods analysed, to process the grey level images has been demonstrated.

The neurally controlled colour patterns of cephalopod molluscs, and the spatial distribution of the spots in the skin that supply the display elements for these patterns, have been followed during different stages of the life history. The resulting description links physiological with morphogenetic processes of pattern generation and provides a novel contribution to understanding some network properties of this and other systems. Upon denervation of one side of the animal, each of the several generations of spots turns out to be a separate ‘neural’ net composed mainly of muscles. Waves of colour propagate through these networks with unstable periodicities and in semi-random directions. They are manifestations of the intrinsic connectivity and conductances of the networks. The role of the nervous system is to control the chaos of myogenic origin by modulating either conduction or coupling in the networks.

“ConscÍousness” is a multiply ambiguous word, and if our goal is to explain perceptual consciousness we had better be clear about which of the many senses of the word we are endorsing when we sign on to the project. I describe some of the relatively standard distinctions made in the philosophical literature about different meanings of the word “conscious”. Then I consider some of the arguments of David Chalmers and of Ned Block that states of “phenomenal consciousness” pose special and intractable problems for the scientific understanding of perception. I argue that many of these problems are introduced by obscurities in the term itself, and propose a distinction between epistemic and non-epistemic senses of the term “phenomenal consciousness”. That distinction helps explain why phenomenal consciousness seems so mysterious to so many people. States of “phenomenal consciousness” are not states of one, elemental (and inexplicable) kind; they are a ragtag lot, of differing levels of complexity, corralled under one heading by a regrettable ambiguity in our terminology.

There seems to be general agreement amongst researchers investigating detailed structures in the brain that concepts like mind and consciousness will ultimately have their explanation in terms of classical physics with quantum mechanics playing no role at these higher levels. However classical physics excludes the observer, whereas the higher level activities of the brain involve the observer in the form of a subject in a very fundamental way. In quantum theory the observer plays a crucial role and there are general arguments which suggest that quantum theory will play a key role at this level. In this paper I will discuss the reasons for adopting such a position.

I will discuss Bohr’s contribution to the mind/matter debate, a contribution which depends on the recognition that there is an element of wholeness in quantum processes that leads to general problems concerning the separation of subject from object. I will show how the work I did with Bohm led to similar conclusions, albeit from a very different point of view. Our view suggests that physical processes contain an organic element that is based on the notion of active information. I also present some more general ideas which replace the Cartesian order by a more subtle order, the implicate order which I feel offers the best possibility of bridging the gap between the basic underlying neurophysics and the top down psychophysics.

Colour vision has a causal chain structure. The chain consists of an electrical neuronal processing part, coding for the amounts of elementary colour sensations (elementary colours), as well as for the spatial locations of the composed colour sensations (colours), and finally of the elementary colour sensations per se. The composed colour sensations consist of six elementary colours, which differ qualitatively from each other. On the other hand, the electrical membrane potentials of the neurones are always one physical quality only, even when superimposed from several neurones and vary over time. This ontological gap in the physiological description of colour vision was, therefore, investigated as a black-box with the electrical excitations of the colour coding (CC) neurones as the input, and the amounts of the elementary colours (EC) as the output. The general properties of a psychophysiological model of colour sensations were theoretically determined, and measured in psychophysical experiments in conjunction with introspection. The results of these investigations are presented and discussed. Spatial colour perception turned out to have in the order of 10¹⁶ colour locations, allowing for hyperacuity judgements. The necessary colour and spatial information can actually be provided by the much smaller number of neurones in the cortical visual areas. The psychophysiological model of colour sensations is expected to finally enable us to identify the materials that are close to or even identical to the elementary colour sensations.

Four terms (red, green, yellow, and blue) are necessary and sufficient to describe the appearance of hue. Parvo cells of the LGN respond maximally to modulation along either an S-(L+M) or an L-M axis, but few cells are tuned to intermediate directions. These axes do not correspond to the colour appearance axes. Colour appearance requires at least one more stage of processing, a stage that transforms the second stage axes in the cortex. Receptive fields in area V l vary much more in their colour tuning than in LGN, and they have significantly more S-cone input. Thus, a major accomplishment of striate cortex may be rotation of the colour axes by integrating S-cone input. Major differences in chromatic properties of receptive fields between area VI and areas V2 and V3 have not been reported. V4 cells, however, have large receptive field surrounds that may support colour constancy and promote segregation of figure and ground by colour. It seems unlikely that V4 is a “colour centre” in the monkey brain, and it is equally unlikely to be homologous with those regions of human cortex associated with achromatopsia.

Training experiments with a series of small test fields ranging in color between yellow and blue, or between green and red allowed the investigation of color constancy in goldfish quantitatively. Color constancy was perfect within certain limits when the illumination was changed from white to yellow, blue, red or green. Color constancy was also found to depend on background lightness and size. It was perfect with grey background even under relatively high saturation levels of illumination color. With a white background and red illumination a color contrast effect was observed, which was interpreted as an overcompensation of the color constancy mechanisms. In training experiments and transfer tests we could show that the goldfish uses color categories under certain conditions. The results are compared with the outcome of generalization experiments which does not seem to reflect categorization but rather discrimination ability.

The bottleneck of the optic nerve demands very efficient, compact coding in the retina, with each neuron in the path carrying the maximum amount of information. With the large number of cells in the cortical visual regions, the requirements and the nature of the processing are quite different. The initial cortical processing involves an unpacking of the retinal information into multiple channels, each being selective for a different subsample of the local information. The result is a sparse coding in which most neurons do not respond at all to any given stimulus, but the particular ones that do respond identify by virtue of their response important characteristics of the stimulus pattern. At the level of the striate cortex (VI), the processing occurs over somewhat larger visual regions than in the lateral geniculate nucleus (LGN), but it is still quite local. The well-established selectivities of the various cells at this level are for the local 2-dimensional spatial frequency (spatial frequency at a particular orientation), the local direction of motion, the local binocular disparity, and the local color.

Most of the neurons that transmit information from the retina to the cortex are quite linear, insuring little loss of information. Neurons in Vl, however, have various significant non-linearities which enhance their selectivity and increase the sparseness of the coding. A second level of processing within the striate cortex also produces complex cells, which have somewhat larger receptive fields within which they respond to patterns independent of the patterns’ spatial phase.

The primary role of the striate cortex appears to be to transform and encode information about the local pattern in ways that make it convenient for cells at later striate levels to use, At successive later levels, cells perform increasingly global analyses, with processing along different stimulus dimensions taking place to some extent in different prestriate regions. A prime initial task for these later levels must be image segmentation into objects: identifying certain parts of the visual field as all belonging together as one object and other parts as being the background or belonging to some different object. This problem was identified and has been extensively studied by Gestalt psychologists and others, but little is known of how it is accomplished neurally. A final, even less well understood process is that of object recognition, which is presumably accomplished by comparing the stimulus-driven activity with stored information about the world.

Kinetic boundaries are interesting subjects of study because of their computational power. They are perceived as precisely in spatial terms as luminance defined gratings but require longer presentation times to become visible. The evidence is reviewed that kinetic and luminance- defined boundaries have converged onto the same neurones by the time their signals reach high levels in the visual system. Whlile it is increasingly clear that area MT/V5 and its human homologue is little involved in processing kinetic contours, the area supposedly pre-processing these velocity fields has been identified only in humans (region KO). The failure to identify the monkey homologue of this region has slowed the progress on the question of exactly at what level in the visual system kinetic and luminance signals initially converge.

The direct mechanical control of ion channels enables each single mechanoreceptor cell to represent in its receptor potential complex sequential sounds (music) in a comparable quality as the “microphonic” summating potential of the total organ of Corti. This was demonstrated for a bristle-receptor of a fly. Two reasons are considered for the involvement of ten thousands of sensory and ganglional cells in the mammalian auditory system: 1). Differentiated acoustic induction of behaviour and conscious experiencing of music are based on distributed representation of processed information (practising a low information channel capacity of the individual neuron), not on physiological representation per se. 2.) Only a very small percentage of the high information content present in the receptor potential of a single cell can be transmitted to other cells.

Adaptation is used by most mechanoreceptor cells to condense the stimulatory information onto the limited range of response amplitudes. Different from the initial rapid process of conductance control, most mechanisms of the slower adaptation involve active processes requiring metabolic energy. Adaptation as studied in a bristle receptor of blowflies strongly depends on intracellular Ca²⁺-activity. Resting sensitivity and stimulus-induced extent and rate of adaptation can be regulated by metabolites of arachidonic acid.

Communication by means of sound is not always easy. Sound suffers much attenuation and degradation close to ground. Crickets have adapted to this by exploiting sharply tuned mechanical systems. As a result, the calling song of males and the directional hearing of females are tuned to the same frequency. The calling song is emitted by a lightly damped part of the front wing, known as the harp. A mechanical feedback between the oscillating harp and the stridulatory apparatus ensures that the harp is driven at its resonance frequency. A uniform carrier frequency is needed, because the directional hearing of listening crickets is also tuned to a narrow frequency range. The cricket obtains sufficient directionality for overcoming the degradation of directional cues in its biotope by exploiting a phase shifter in the tracheal tube guiding sound through the body to the contralateral ear. By means of the phase shifting mechanism, the cricket can obtain a proper phase relationship between the sounds acting on its ear drums, but only within a narrow range around the frequency of the conspecific calling song.

A short review is given on the olfactory and gustatory systems in vertebrates, and mostly in mammals (including humans) with special reference to mechanisms taking part in sensory coding.

Insects detect small changes in odour concentration at astoundingly low thresholds and recognize and distinguish complex blends of odours. Furthermore, they are able to localize odour sources within short range as well as over long distances. Specialized olfactory receptor neurons on the antennae encode the different properties of odours in their action potential response. Via the axonal projections of olfactory receptor neurons into the antennal lobes the olfactory information is relayed into the deutocerebrum of the midbrain. From the antennal lobes the olfactory information is distributed into protocerebral target areas such as the mushroom bodies and the lateral protocerebrum. What kind of information is encoded in these midbrain regions and how these areas finally connect to locomotor control centers remains largely unknown. Therefore, this paper will concentrate on olfactory information processing in the antennal lobes of moths as the best studied system.

While foraging, Cataglyphis ants of the Saharan desert leave their underground burrows individually for distances of several hundred metres in circuitous ways, but then return along the bee-line - or ant-line, so to speak - directly to the starting point. Behavioural experiments show that they accomplish this task by path integration (dead reckoning). In mathematical terms, path integration requires trigonometric computations. Cataglyphis, however, lives within a Pre-Pythagorean world. It dissects the problem into a number of subroutines by using multiple mechanisms, or behavioural modules, which act in concert. Path integration, however, is an intrinsicly noisy system. Based on an egocentric system of reference it is prone to cumulative errors. Supplementary corrective inputs from external guides are necessary. And landmarks are such guides. In fact, they provide the animal with snapshot memories and landmarkbased local vectors. The results suggest that the insect exhibits complex multi-layered memory stores, from which information is retrieved in flexible, context-dependent ways. Moreover, they emphasize the point that the insect’s strategies are well adapted to the navigator’s real-time and real-life performances, and that in the course of evolution Cataglyphis might have neither acquired the capacity nor encountered the need for more elaborate all-purpose systems of navigation.

Avian navigation consists of a two steps process: birds first determine the direction to the goal as a compass course, then they locate this course with the help of a compass. For both steps, birds can utilize a variety of cues. The magnetic compass plays a central role in the navigational system, as it provides young birds with a first mechanism for directional orientation and navigation. At the same time, it allows them to calibrate other types of cues and to establish the complex, experiencedependent mechanisms preferentially used by experienced birds. The sun compass is established during a sensitive period by combining sun azimuth, time of day as indicated by the internal clock and direction as indicated by the magnetic compass. The navigational ‘map’ is established by combining the net direction of the outward journey with site-specific information involving magnetic parameters, odours, infrasound, landscape features etc. The general principle for the development of the avian navigational system thus is to use a simple innate mechanism - the magnetic compass - as a basis for learning processes that form complex, multifactorial mechanisms utilizing celestial and/or geophysical and other environmental cues.

The correlations have been analyzed between the impairments of higher cortical functions (assessed by clinical scores and neuropsychological tests) and multichannel EEG spectra in both elderly patients suffered with “mild” dementia, and children with cognitive problems. The decrease of regional cerebral blood flow, disbalance between neocortical and limbic structures, and lack of inhibitory influences from the frontal cortex are discussed as tentative causes of these brain dysfunctons associated with the “slow sensorimotor rhythm”.

In human subjects lesions restricted to the frontal lobes cause profound changes in personality and intellect. The core-defect of “frontal-lobe patients” consists in a general loss of behavioural guidance and organizational deficits. Animal experiments have shown that the prefrontal cortex contains distinct classes of neurons which are engaged in registering, internal processing and responding to a stimulus respectively. Human studies with brain damaged patients as well as functional neuroimaging studies in normal controls have evidenced that different areas of the prefrontal cortex make distinct contributions to cognitive processes, such as memory. Thus, there is converging evidence from several lines of study pointing to the eminent role of the frontal lobes in the orchestration of complex behaviour. This contribution provides an overview with a particular focus on the role of the prefrontal cortex in memory processing.

When dealing with perceptual systems our attention may span from the side of the neuronal activity linked to sensory information processing and neuronal coding related to the perception event, to the side of the consciousness content associated with perception in humans. While blindsight, subliminal perception, and other cases of nonphenomenal seeing and related phenomena speak contrary to the absolute connection between perception and consciousness in humans, the absolute nonconnection has been claimed for perception in nonhuman animals since the behaviorist age. An attitude more open towards the possible consciousness content of nonhuman perception is solicited, in spite of undeniable difficulty in treating this topic.

The following sections are included:

• PARTICIPANTS