Magnetophosphenes: a quantitative analysis of thresholds

Low-frequency and transient magnetic fields of moderate flux densities are known to generate visual phenomena, so-called magnetophosphenes. In the present study, time-variable very low frequency (10–50 Hz) electromagnetic fields of moderate flux density (0–40 mT) were used to induce magnetophosphenes. The threshold values for these phosphenes were determined as a function of the frequency of the magnetic field both in normal subjects and colour defective ones. Maximum sensitivity occurred at a frequency of approximately 20–30 Hz, and with broad-spectrum light the threshold flux density was 10–12 mT. The threshola values were found to be dependent upon the intensity and the spectral distribution of the background light. Sensitivity decreased during dark adaptation. In certain respects deutans differed from subjects with normal colour vision. Possible mechanisms for generation of magnetophosphenes are discussed. The present magnetic threshold curves show a close resemblance to corresponding curves obtained by electric stimulation at various frequencies provided the electric thresholds are divided by the a.c. frequency. These problems are under current investigation in our laboratory. This is in full agreement with the assumption that the fluctuating magnetic field affects retinal neurons by inducing currents which polarise synaptic terminals.     

Computational analysis of thresholds for magnetophosphenes

In international guidelines, basic restriction limits on the exposure of humans to low-frequency magnetic and electric fields are set with the objective of preventing the generation of phosphenes, visual sensations of flashing light not caused by light. Measured data on magnetophosphenes, i.e. phosphenes caused by a magnetically induced electric field on the retina, are available from volunteer studies. However, there is no simple way for determining the retinal threshold electric field or current density from the measured threshold magnetic flux density. In this study, the experimental field configuration of a previous study, in which phosphenes were generated in volunteers by exposing their heads to a magnetic field between the poles of an electromagnet, is computationally reproduced. The finite-element method is used for determining the induced electric field and current in five different MRI-based anatomical models of the head. The direction of the induced current density on the retina is dominantly radial to the eyeball, and the maximum induced current density is observed at the superior and inferior sides of the retina, which agrees with literature data on the location of magnetophosphenes at the periphery of the visual field. On the basis of computed data, the macroscopic retinal threshold current density for phosphenes at 20?Hz can be estimated as 10?mA m(-2) (-20%?to + 30%, depending on the anatomical model); this current density corresponds to an induced eddy current of 14 μA (-20%?to + 10%), and about 20% of this eddy current flows through each eye. The ICNIRP basic restriction limit for the induced electric field in the case of occupational exposure is not exceeded until the magnetic flux density is about two to three times the measured threshold for magnetophosphenes, so the basic restriction limit does not seem to be conservative. However, the reasons for the non-conservativeness are purely technical: removal of the highest 1% of electric field values by taking the 99th percentile as recommended by the ICNIRP leads to the underestimation of the induced electric field, and there are difficulties in applying the basic restriction limit for the retinal electric field.     

Electrical phosphenes: on the influence of conductivity inhomogeneities and small-scale structures of the orbita on the current density threshold of excitation

Electrical and magnetic phosphenes, perceptions of light as a result of non-adequate stimulation of the eye by electrical current or magnetic induction, respectively, are one of the cornerstones to justify limit values for extreme low-frequency fields specified by statutory regulations. However, the mechanism and place of action, as well as the excitation threshold, remain unknown until now. We suggest that the origin of phosphene excitation is the synaptic layer of the eye. The current density threshold value for electrical phosphene excitation was numerically quantified for this area on the basis of a detailed geometrical model in original submillimetre resolution and specifically measured conductivities in the LF range. The threshold values found were 1.8 Am⁻² at 60 Hz and 0.3 Am⁻² at 25 Hz. These values are comparable with values of other excitable tissues. It has been shown that the current density threshold for phosphene generation depends on small-scale structures not taken into account by previous models.     

Microstimulation of V1 delays visually guided saccades: a parametric evaluation of delay fields

Electrical microstimulation of macaque striate cortex (area V1) delays the execution of saccadic eye movements made to a visual target placed in the receptive field of the stimulated neurons. The region of visual space within which saccades are delayed is called a delay field. We examined the effects of changing the parameters of stimulation and target size on the size of a delay field. Rhesus monkeys were required to generate a saccadic eye movement to a punctate and white visual target presented within or outside the receptive field of the neurons under study. On 50% of trials, a train of stimulation consisting of 0.2-ms anode-first pulses was delivered to the neurons before the onset of the visual target. Stimulations were performed in the operculum at 0.9–2.0 mm below the cortical surface. It was found that increases in current (50–100 μA), pulse frequency (100–300 Hz), or train duration (75–300 ms) increased the size of a delay field and increases in target size (0.1°–0.2° of visual angle) decreased the size of a delay field. Delay fields varied in size between 0.1 and 0.6° of visual angle. These results are related to the properties of phosphenes induced by electrical stimulation of V1 in humans and compared to the interference effects observed following transcranial magnetic stimulation of human V1.     

Modulation of visual cortical excitability in migraine with aura: effects of 1 Hz repetitive transcranial magnetic stimulation

Recent studies showed hyperexcitability of the occipital cortex in subjects affected by migraine with aura. It has been shown that 1 Hz repetitive transcranial magnetic stimulation (rTMS) reduces excitability of visual cortex in normal subjects. The aim of the study was to investigate the effects of low frequency (1 Hz) rTMS on visual cortical excitability by measuring changes in phosphene threshold (PT) in subjects with migraine with aura. Thirteen patients with migraine with aura and 15 healthy controls were examined. Using a standardized transcranial magnetic stimulation protocol of the occipital cortex, we assessed the PT (the lowest magnetic stimulation intensity at which subjects just perceived phosphenes) before and after a 1-Hz rTMS train delivered at PT intensity for 15 min. The difference in the proportion of subjects reporting phosphenes in migrainer and control groups was significant (migrainers: 100% vs controls 47%; P<0.05), and 1 Hz rTMS over the occipital cortex led to a significantly increased visual cortex excitability expressed as a decrease in PT in subjects affected by migraine with aura. Conversely, after a 1-Hz TMS train normal subjects showed increased PT values, which suggests a decreased visual cortex excitability. Our findings confirm that the visual cortex is hyperexcitable in migrainers and suggest a failure of inhibitory circuits, which are unable to be upregulated by low frequency rTMS.     

Modulatory effects of low- and high-frequency repetitive transcranial magnetic stimulation on visual cortex of healthy subjects undergoing light deprivation

The aim of the present study was to explore further the effects of light deprivation (LD) on visual cortex excitability. Healthy subjects reporting reliable induction of phosphenes by occipital transcranial magnetic stimulation (TMS) underwent 60 min of complete LD. Phosphene threshold (PT) was measured before ( T ₀), after 45 min ( T ₁) and 60 min ( T ₂) of LD, and then every 10 min after light re-exposure until recovery to T ₀ values. Repetitive TMS (rTMS) (at 1 or 10 Hz) was applied in separate sessions during the last 15 min of LD. PTs significantly decreased after 45 min of LD. rTMS differentially modified the effects of 60 min LD on PTs depending on stimulation frequency. One hertz rTMS did not change the decreasing of PT values as observed in baseline condition, but significantly prolonged the time to recover T ₀ PT values after light re-exposure. By contrast, 10 Hz rTMS significantly increased PT and the time to recover T ₀ PT values after light re-exposure was shortened. The results of this study show that the modulatory effects of different rTMS frequencies on visual cortex critically depend on the pre-existing excitability state of inhibitory and facilitatory circuits, and provide novel insights into the neurophysiological changes that take place in the visual cortex following functional visual deafferentation.     

Influence on frog retina of alternating magnetic fields with special reference to ganglion cell activity

The effects of extremely low frequency (e.l.f.) electromagnetic fields on basal systems were studied. Frog retinas were exposed to magnetic fields with frequencies and flux densities that have been shown to induce magnetophosphenes in volunteers. The electrical activity in the retina induced by the field was recorded from the ganglion cell layer with a microelectrode technique. A threshold value was obtained at approximately 20 mT, and a sensitivity maximum at 20 Hz. A significant prolongation (4 ms) of the latency from light stimulus to response in the ganglion cell layer was obtained, if the preparation was simultaneously and continuously exposed to a magnetic field. A study of the reaction of the ganglion cells to light and to magnetic fields showed that those cells which were on-cells during light stimulation became off-cells during magnetic stimulation andvice versa. The magnetic field response occurred within approximately 5 ms. while the light stimulus response occurred only after an average of approximately 85 ms. Addition of Na-aspartate or CoCl₂ extinguished simultaneously the response both to light and to magnetic field stimuli.     

A model of the electrical volume conductor in the region of the eye in the ELF range.

Electrical and magnetic phosphenes are irritations of the eye caused by electric currents or magnetic fields. These are well known effects initially investigated in the early 1900s. Available estimations of the current densities in the eye, based on the assumption of a homogeneous volume conductor, show low thresholds. These outdated thresholds are still an important cornerstone when justifying today's limit values for extremely low-frequency (ELF) fields specified by statutory regulations. In vitro measurements of the complex conductivity of cattle eye are carried out for the ELF range (5-2000 Hz) separated for the different tissues of the eyeball. They do not show peculiarities at 20 Hz which is the threshold minimum for the phosphene generation. The reported conductivity data of the eye region show variations of two orders of magnitude regarding the electrical conductivity of the individual tissue layers. Starting with these new data, a model of the orbita is introduced describing the eye and its periphery as an electrically inhomogeneous volume conductor. This model contains small-scale structures which are expected to behave as good electrical conductors yielding regions of higher field values within the eye. Therefore, earlier models assuming a homogeneous volume conductor can be regarded as oversimplistic.     

Estimation of the distribution of intramuscular current during electrical stimulation of the quadriceps muscle

Electrical stimulation is commonly used for strengthening muscle but little evidence exists as to the optimal electrode size, waveform, or frequency to apply. Three male and three female subjects (22–40 years old) were examined during electrical stimulation of the quadriceps muscle. Two self adhesive electrode sizes were examined, 2 cm × 2 cm and 2 cm × 4 cm. Electrical stimulation was applied with square and sine waveforms, currents of 5, 10 and 15 mA, and pulse widths of 100–500 μs above the quadriceps muscle. Frequencies of stimulation were 20, 30, and 50 Hz. Current on the skin above the quadriceps muscle was measured with surface electrodes at five positions and at three positions with needle electrodes in the same muscle. Altering pulse width in the range of 100–500 μs, the frequency over a range of 20–50 Hz, or current from 5 to 15 mA had no effect on current dispersion either in the skin or within muscle. In contrast, the distance separating the electrodes caused large changes in current dispersion on the skin or into muscle. The most significant finding in the present investigation was that, while on the surface of the skin current dispersion was not different between sine and square wave stimulation, significantly more current was transferred deep in the muscle with sine versus square wave stimulation. The use of sine wave stimulation with electrode separation distances of less then 15 cm is recommended for electrical stimulation with a sine wave to achieve deep muscle stimulation.     

Optimization of single electrode tactile codes

The effects of a frequency modulated electrocutaneous signal's (code's) characteristics on the interpretability of the signal were investigated using an electrocutaneous tracking approach. The characteristics investigated include the functional relationship (exponential and hybrid) between an informational signal and the stimulation frequency, the range of stimulation (2–50 Hz and 2–100 Hz), and the impact of pulse width compensation on a code's efficacy. The interpretability of six different single bipolar electrode codes was examined by 30 subjects using a balanced incomplete block experimental design. Codes with exponentially shaped transfer functions resulted in generally lower electrocutaneous tracking errors than codes utilizing hybridshaped transfer functions. Hybrid codes had a transfer function that was linear in the lower frequency range and exponential in the higher frequency range. Codes with a 2–100 Hz frequency range were interpreted better than codes with a 2–50 Hz frequency range. The use of pulse width compensation to maintain a more even level of stimulation intensity had a slightly negative effect on the subjects' abilities to cutaneously track the information signal.     

Hazardous nature of high-temporal-frequency strobe light stimulation: neural mechanisms revealed by magnetoencephalography

Individuals in contemporary society are continually exposed to various visual stimuli. Such stimulation, especially when high in temporal frequency, may sometimes cause unexpected events such as photosensitive seizures. Although many studies have demonstrated that high-temporal-frequency (>3 Hz) visual stimulation can yield hazardous responses in the CNS, the mechanisms by which it does so are still unclear. We therefore investigated the mechanisms of neural perturbation by high-temporal-frequency strobe light stimulation with high-temporal-frequency resolution (4–20 Hz with an interval of 2 Hz) using magnetoencephalography with high temporal and spatial resolution. We show that (1) three temporal dipole phases (phases 1, 2 and 3, by time course) can be identified in the visual evoked magnetic fields (VEF's) across stimulation frequencies based on the goodness-of-fit values for equivalent current dipole estimation and horizontal dipole directions, (2) the dipole moment of VEF's is correlated with autonomic nervous system activity in phases 1 and 2, (3) some temporal stimulation frequencies enhance magnetic responses in phases 1, 2 and 3, and (4) these frequencies are harmonically related, with a greatest common divisor frequency (fundamental frequency) of approximately 6.5 Hz. Our clarification of the temporal frequency characteristics of VEF's will contribute to understanding of the potential hazardous effects of high-temporal-frequency strobe light stimulation in the CNS.     

Cortical Activity Prior to, and During, Observation and Execution of Sequential Finger Movements

Summary The aim of this study was to provide further evidence for the existence of a mirror neuron system in humans using electroencephalography during the observation and execution of non-object-related movements. Event-related desynchronization and synchronization (ERD/ERS) were used to characterize brain activity prior to, and during, observation and execution of a finger movement in four frequency bands (7–10 Hz, 10–13 Hz, 13–20 Hz, and 20–30 Hz). Electroencephalograms (EEGs) were recorded from 19 electrode sites in eight participants. In all the frequency bands and electrode sites, results revealed that there was no significant differences in EEG cortical activity between the observation condition and the execution conditions. Comparison of the two stages of the movement (i.e., pre-movement and movement) in the observation and execution conditions showed, in most cases, that pre-movement ERD values were less than movement ERD values. Whilst there was not an identical match of EEG cortical indices, this study provides further support for the existence of a mirror neuron system in humans. The incomplete congruence may be explained by the different behaviors, the nature of the task and factors in the observed action coded by the mirror system.     

Histologic and physiologic evaluation of electrically stimulated peripheral nerve: Considerations for the selection of parameters

Helical electrodes were implanted around the left and right common peroneal nerves of cats. Three weeks after implantation one nerve was stimulated for 4–16 hours using charge-balanced, biphasic, constant current pulses. Compound action potentials (CAP) evoked by the stimulus were recorded from over the cauda equina before, during and after the stimulation. Light and electron microscopy evaluations were conducted at various times following the stimulation. The mere presence of the electrode invariably resulted in thickened epineurium and in some cases increased peripheral endoneurial connective tissue beneath the electrodes. Physiologic changes during stimulation included elevation of the electrical threshold of the large axons in the nerve. This was reversed within one week after stimulation at a frequency of 20 Hz, but often was not reversed following stimulation at 50–100 Hz. Continuous stimulation at 50 Hz for 8–16 hours at 400 μA or more resulted in neural damage characterized by endoneurial edema beginning within 48 hours after stimulation, and early axonal degeneration (EAD) of the large myelinated fibers, beginning by 1 week after stimulation. Neural damage due to electrical stimulation was decreased or abolished by reduction of the duration of stimulation, by stimulating at 20 Hz (vs. 50 Hz) or by use of an intermittent duty cycle. These results demonstrate that axons in peripheral nerves can be irreversely damaged by 8–16 hours of continuous stimulation at 50 Hz. However, the extent to which these axons may subsequently regenerate is uncertain. Therefore, protocols for functional electrical stimulation in human patients probably should be evaluated individually in animal studies.     

Closed-chest cardiac stimulation with a pulsed magnetic field

Magnetic stimulators, used medically, generate intense rapidly changing magnetic fields, capable of stimulating nerves. Advanced magnetic resonance imaging systems employ stronger and more rapidly changing gradient fields thant those used previously. The risk of provoking cardiac arrhythmias by these new devices is of concern. In the paper, the threshold for cardiac stimulation by an externally-applied magnetic field is determined for 11 anaesthetised dogs. Two coplanar coils provide the pulsed magnetic field. An average energy of approximately 12kJ is required to achieve closed-chest magnetically induced ectopic beats in the 17–26kg dogs. The mean peak induced electric field for threshold stimulation is 213 Vm ⁻¹ for a 571 μs damped sine wave pulse. Accounting for waveform efficacy and extrapolating to long-duration pulses, a threshold induced electric field strength of approximately 30 Vm ⁻¹ for the rectangular pulse is predicted. It is now possible to establish the margin of safety for devices that use pulsed magnetic fields and to design therapeutic devices employing magnetic fields to stimulate the heart.     

Effect of skin temperature on vibrotactile sensitivity

The vibration problems relating to living bodies have so far been studied from the perspectives of engineering physiology and psychology. This study shows the relationship between vibratory sensibility and temperature in the living body. Psychological experiments were carried out by using the vibrometer of an acoustic calibration apparatus in sine, triangular and square waves. The sensibility-threshold measurements were made using 30–700 Hz sine waves, 30–300 Hz triangular and sawtooth waves, or 30–250 Hz square waves. Each of ten subjects was kept seated. The average value of the vibratory levels, varied by ascending and descending steps, was taken as that of the threshold. As the vibrometer in the apparatus used makes a noise at frequencies greater than 250 Hz it was masked from the subject by presenting him with a different noise. The threshold curve for square waves is lower by 12·3 dB than that for sine waves at about 30Hz. The threshold curve of the 26°C sine wave was lower by 10 dB than that of the 58°C sine wave vibration near 200 Hz. For example with a sine wave, at 58°C the amplitude threshold was lowest at about 270 Hz, but at −11°C at about 200 Hz. At frequency stimulation higher than 120 Hz, as the temperature of the contact point was lowered, the amplitude threshold increased and the frequency at which the threshold curve was at a minimum shifted to a lower frequency.     

Influence of stimulation of the medial hypothalamus on the interaction of neurons of the rabbit neocortex

The interaction of neurons of the visual and sensory motor areas of the neocortex of the rabbit before and after stimulation of some medial nuclei of the hypothalamus was investigated by plotting cross- and autocorrelation histograms. Stimulation through bipolar electrodes using bursts of biphasic pulses at a frequency of 100 Hz, current strength 50–200 μA, led to the appearance in freely behaving rabbits of the reaction of avoidance of the place of stimulation. Following stimulation, as compared with resting wakefulness, the number of pairs of neurons functioning in correlation increased to 45%; at the same time, discharges of neurons of the sensory motor area ran ahead of discharges of visual neurons in the pairs up to 120 msec; the periodicity of the coupled discharges was mainly in the theta frequency range. A conclusion regarding the reflection of defense motivation in certain indices of the interaction of the cortical cells in the presence of a tonic conditioned reflex is reached on the basis of a comparison of the interaction of neurons following stimulation of the medial hypothalamus and the midbrain reticular formation, in the intersignal periods during the development of a defense conditioned reflex as well. This study was supported by the Russian Basic Research Fund (project No. 94-04-11399 a). Institute of Higher Nervous Activity and Neurophysiology, Russian Academy of Sciences, Moscow. Translated from Zhurnal Vysshei Nervnoi Deyatel'nosti imeni I. P. Pavlova, Vol. 45, No. 2, pp. 297–304, March–April, 1995.     

Effects of high-rate electrical stimulation upon firing in modelled and real neurons

Many medical devices use high-rate, low-amplitude currents to affect neural function. This study examined the effect of stimulation rate upon action potential threshold and sustained firing rate for two model neurons, the rabbit myelinated fibre and the unmyelinated leech touch sensory cell. These model neurons were constructed with the NEURON simulator from electrophysiological data. Alternating-phase current pulses (0–1250 Hz), of fixed phase duration (0.2 ms), were used to stimulate the neurons, and propagation success or failure was measured. One effect of the high pulse rates was to cause a net depolarisation, and this was verified by the relief of action potential conduction block by 500 Hz extracellular stimulation in leech neurons. The models also predicted that the neurons would maintain maximum sustained firing at a number of different stimulation rates. For example, at twice threshold, the myelinated model followed the stimulus up to 500 Hz stimulation, half the stimulus rate up to 850 Hz stimulation, and it did not fire at 1250 Hz stimulation. By contrast, the unmyelinated neuron model had a lower maximum firing rate of 190 Hz, and this rate was obtained at a number of stimulation rates, up to 1250 Hz. The myelinated model also predicted sustained firing with 1240 Hz stimulation at threshold corrent, but no firing when the current level was doubled. Most of these effects are explained by the interaction of stimulus pulses with the cell's refractory period.     

Short and medium latency muscle responses evoked by electrical vestibular stimulation are a composite of all stimulus frequencies

Electrical vestibular stimulation produces biphasic responses in muscles maintaining balance. The two components of these muscle responses (termed the short latency and medium latency components) are believed to be independent and elicited by vestibular stimuli of different frequencies. We tested these hypotheses by determining (a) if frequency-specific stimulation protocols could evoke independently the short and medium latency responses and (b) whether these two components are triggered by distinct brain regions with a fixed time delay, interacting around 10 Hz. First, subjects were provided 10–25 Hz, 0–10 Hz, and 0–25 Hz vestibular stimuli to selectively modulate the short latency, medium latency, or both components of the response; and second, they were provided twenty sinusoidal stimuli from 1 to 20 Hz with a 0–20 Hz control trial, designed to determine whether an interaction between the short and medium latency responses occurs at a specific stimulation frequency. Both the 0–10 Hz and 10–25 Hz vestibular stimuli elicited multiphasic waveforms, suggesting the short and medium latency components were not modulated independently by the frequency-specific stimuli. Sinusoidal vestibular stimuli evoked responses at the stimulated frequency but no evidence of a reflex component interaction was observed. Instead, summation of the responses evoked by each of the sinusoidal stimuli resembled the biphasic response to broad bandwidth stimuli. Due to the lack of interaction and linear contribution of all stimulus frequencies to both the short and medium latency responses, the present results support the use of broad bandwidth electrical vestibular signal for physiological or clinical testing.     

Experimental muscle pain decreases the frequency threshold of electrically elicited muscle cramps

This study in humans tested the hypothesis that nociceptive muscle afferent input facilitates the occurrence of muscle cramps. In 13 healthy adults, muscle cramps were experimentally induced in the foot by stimulating the tibialis posterior nerve at the ankle with 2-s bursts of stimuli separated by 30 s, with stimulation frequency increasing by 2-Hz increments from 10 Hz until the cramp appeared. The minimum stimulation frequency that induced the cramp was defined “cramp frequency threshold”. In 2 days, elicitation of the cramp was performed in the two-feet with and without (baseline condition) injection of hypertonic (painful condition) or isotonic (control condition) saline into the deep midportion of the flexor hallucis brevis muscle, from where surface EMG signals were recorded. The cramp frequency threshold was lower for the painful condition with respect to its baseline (mean ± SE, hypertonic saline: 25.7 ± 2.1 Hz, corresponding baseline: 31.2 ± 2.8 Hz; P < 0.01) while there was no difference between the threshold with isotonic injection with respect to baseline. EMG average rectified value and power spectral frequency were higher during the cramp than immediately before the stimulation that elicited the cramp (pre-cramp: 13.9 ± 1.6 μV and 75.4 ± 3.8 Hz, respectively; post-cramp: 19.9 ± 3.2 μV and 101.6 ± 6.0 Hz; P < 0.05). The results suggest that nociceptive muscle afferent activity induced by injection of hypertonic saline facilitates the generation of electrically elicited muscle cramps.     

Transient visual field defects induced by transcranial magnetic stimulation over human occipital pole

Transient visual field defects (VFDs) and phosphenes were induced in normal volunteers by means of transcranial magnetic stimulation (TMS) using a circular magnetic coil of 12.5 cm diameter placed with its lower rim 2–4 cm above the inion in the midline. Subjects had to detect small, bright dots presented randomly for 14 ms in one of 60 locations on a computer screen resulting in a plot of the central 9° of the visual field. In 8 of 17 subjects, transient VFDs were inducible at peak magnetic field strenghts of 1.1–1.4 T. In the central 1–3°, detection of targets was impaired in both the upper and lower visual field, whereas at 4–9° large parts of only the lower visual field were affected with a sharp cut-off along the horizontal meridian. Targets at 1° in the lower field were affected with lower TMS intensities than corresponding locations in the upper or peripheral locations in the lower field. Detection of central targets was affected at more caudal stimulation sites than detection of peripheral targets. Phosphenes were elicitable in 14 of 17 subjects at clearly lower field strengths of 0.6–1.0 T. Many subjects perceived chromatophosphenes. From a discussion of the literature on patients with VFDs and the known topography of the human visual system, it is concluded that the transient VFDs at 1–3° are probably due to stimulation of both striate cortex (V1) and extrastriate areas (V2/V3), while VFDs in the lower visual field at eccentricities 4–9° are due to stimulation of V2/V3 but not V1. Received: 14 January 1997 / Accepted: 2 June 1997