Steven M. Miller*, Guang B. Liu*, Trung T. Ngo*, Greg Hooper?, Stephan Riek?, Richard G. Carson?, John D. Pettigrew*
Addresses: *Vision Touch and Hearing Research Centre and Department
of Physiology and Pharmacology, ?Cognitive Psychophysiology
Laboratory, and ?Perception and Motor Systems Laboratory, The
University of Queensland, Brisbane, 4072, Australia.
Running Head: Interhemispheric switching and rivalry
Words: 248 (abstract), 3545 (text), 1034 (methods), 1659 (figure captions)
Background: Over 200 years since the first reports of binocular rivalry, Logothetis and his colleagues have challenged suggestions that the phenomenon occurs early in the visual pathway. In alert monkeys, they have shown that neurones in primary visual cortex (V1) continue to respond to their preferred stimulus despite the monkey reporting its absence. Moreover, they found that neural activity higher in the visual pathway is highly correlated with the monkey's reported percept. These and other findings suggest that the neural substrate of binocular rivalry must involve high levels, perhaps the same levels involved in reversible figure alternations.
Results: We present evidence that activation or disruption of a single hemisphere in human subjects affects the perceptual alternations of binocular rivalry. Unilateral caloric vestibular stimulation changes the ratio of time spent in each competing perceptual state. Transcranial magnetic stimulation applied to one hemisphere disrupts normal perceptual alternations when the stimulation is timed to occur at one phase of the perceptual switch, but not at the other. Furthermore, activation of a single hemisphere by caloric stimulation affects the perceptual alternations of a reversible figure, the Necker cube.
Conclusions: Our findings suggest that interhemispheric switching mediates perceptual rivalry. Thus competition for awareness in both binocular rivalry and reversible figures occurs between, rather than within, each hemisphere. This interhemispheric switch hypothesis has implications for understanding the neural mechanisms of conscious experience and also has clinical relevance since the rate of both types of perceptual rivalry is slow in bipolar disorder (manic depression).
Binocular rivalry refers to the alternating perceptual states that occur when different images, such as orthogonal contours, are presented simultaneously, one to each eye . For each presented image, periods of perceptual dominance alternate with periods of perceptual suppression, usually every few seconds. Until recently, this phenomenon was thought to result from reciprocal inhibition between monocular neurones (i.e. neurones responsive to input from only one eye) in separate channels in primary visual cortex (V1) . This model of binocular rivalry however, has been challenged by the single-unit studies of Leopold and Logothetis  which show that only a small percentage of neurones in V1 exhibit activity that is correlated with a monkeyís perceptual reports during rivalry. Moreover, of these neurones, all but one were binocular (i.e. responsive to input from either eye). Sheinberg and Logothetis have further demonstrated that high in the visual pathway, in inferotemporal cortex and the superior temporal sulcus, around 90% of neurones demonstrate activity that is correlated with the perception of an effective visual stimulus [4; see reviews 5,6].
Psychophysical studies are also inconsistent with the monocular channel competition model of binocular rivalry. Kovacs et al.  used a patchwork rivalry paradigm in which one eye was presented with patches of a monkey image interspersed with patches of a jungle scene, while the other eye was presented with the opposite composite pattern. The observers nevertheless reported alternations between the coherent monkey image and the coherent jungle scene. This phenomenon was first demonstrated by Diaz-Caneja in 1928  whose finding was recently replicated and quantified . Such experiments show that the brain can organize aspects of each eyeís presented image into rivalling coherent images. This synthetic capacity during binocular rivalry cannot easily be explained in terms of reciprocal inhibition between monocular channels.
Other psychophysical studies also support the notion that binocular rivalry occurs between neural representations at a high level in the visual pathway rather than between each eyeís presentation. Logothetis et al.  rapidly swapped each eyeís presented image at a rate of 3 Hz and demonstrated that this does not induce rapidly changing perceptual alternations but rather, smooth and slow alternations indistinguishable from normal rivalry. Moreover, the phenomenon of monocular rivalry [11-13] is difficult to explain using monocular channel competition models. When two differently coloured orthogonal gratings are superimposed in the same eye, perception of each grating rivals in a manner similar to binocular rivalry .
In accordance with these psychophysical and single-unit studies, two recent fMRI studies of humans undergoing binocular rivalry have demonstrated brain activation in regions of the visual processing hierarchy beyond V1 [14,15]. Similar high-level and widespread activation patterns during rivalry were recently demonstrated using magnetoencephalography (MEG) . While it is important to understand at what level in the visual pathway binocular rivalry is occurring, there is also a need for specific models of its neural mechanism. It has been suggested that the perceptual alternations in binocular rivalry and reversible figures such as the Necker cube, are the result of modulation of visual processing regions by right-sided fronto-parietal brain regions associated with selective attention and the generation of behaviour [14,17]. This proposal is based on the finding of asymmetric cortical activation in right fronto-parietal cortex during binocular rivalry  and on current understanding of the functions of these regions [14,17].
Here we propose a hypothesis for the neural mechanism of perceptual rivalry that extends this recent evidence that rivalry is a high-level process. We suggest an interhemispheric switch model in which one cerebral hemisphereís high-level visual processing regions adopt one of the rivalling percepts while the other hemisphere adopts the other percept. Competition for awareness during rivalry is therefore occurring between, rather than within, each hemisphereís higher visual regions. This interhemispheric switch hypothesis is based on a number of considerations.
Neuropsychological studies with normal and split-brain subjects support the notions of hemispheric independence and dynamic modularity [18,19], and patients who have had an entire hemisphere surgically removed can sustain a coherent visual percept. The antithetical cognitive styles and moods that have been linked to opposite hemispheric sites might require a mechanism to alternate hemispheric activation [20,21]. Evidence for such hemispheric alternations in humans can be found in the literature on ultradian rhythms of cerebral dominance  (but a periodicity in minutes-hours is indicated rather than the seconds-long periods seen in binocular rivalry). Interhemispheric switching is also evident in birdsong production . Finally, a brainstem-mediated, interhemispheric oculomotor alternation exists in fish  and may have a counterpart in humans with damage to the cerebellum or brainstem .
To test our interhemispheric switch hypothesis of binocular rivalry, we first examined the effect of caloric vestibular stimulation on the perception of rivalling vertical and horizontal drifting gratings. PET  and fMRI  studies have shown that caloric stimulation causes activation in contralateral hemispheric structures that are known to be involved in attentional processing  and binocular rivalry  (e.g. temporo-parietal, insular and anterior cingulate cortex). In a clinical context, this technique can temporarily ameliorate left-sided neglect and anosognosia (denial of disease) associated with right hemisphere damage [20,29]. This ability of caloric stimulation to unilaterally activate the same hemispheric structures implicated in attentional processing and binocular rivalry suggests that, if rivalry is mediated by interhemispheric switching, caloric stimulation should alter the baseline perceptual predominance of one image relative to the other (figure 1). Within-hemisphere competition at any level does not predict an effect from such unilateral hemisphere activation.
Figure 1 Set-up for caloric stimulation and binocular rivalry
experiments and the effects on perceptual predominance predicted by the
interhemispheric switch hypothesis. (a) The rivalry set-up shows
a right-drifting vertical grating being presented to the left eye and an
upward-drifting horizontal grating being presented to the right eye using
liquid crystal shutters to restrict the presentation of each image to its
intended eye. The orthogonal gratings induce binocular rivalry and subjects
report their perceptual alternations using response keys on a keyboard.
The caloric stimulation procedure involves irrigating the external ear
canal with iced water until subjects report vertigo and examiners observe
nystagmus. The stimulation acts via the semicircular canals and brainstem
and results in activation of contralateral structures known to be involved
in attentional processing and binocular rivalry. (b,c) The expected
effects on rivalry alternations from unilateral hemisphere activation (according
to the interhemispheric switch hypothesis) are depicted by theoretical
frequency histograms. These represent the frequency (y-axis) of horizontal
and vertical perceptual intervals in seconds (x-axis) during the rivalry-viewing
period. In (b), there is no baseline predominance of either horizontal
or vertical percepts so unilateral hemisphere activation might be expected
to induce either a horizontal (bottom left) or vertical (bottom right)
predominance. In (c), there is a baseline predominance of the horizontal
percept that might be expected to disappear (bottom left), or even reverse
to a vertical predominance (bottom right) following unilateral hemisphere
activation by caloric stimulation. Actual rather than theoretical frequency
histograms are shown in figure 4.
We next tested predictions that binocular rivalry occurs at the same level as reversible figures, by assessing the effect of caloric stimulation during viewing of the Necker cube, a line diagram with ambiguous perspectives. Similar effects of caloric stimulation on binocular rivalry and Necker cube alternations would be further support for the notion that these phenomena have a common neural mechanism [6,17,30]. If unilateral hemisphere activation induces a change in the baseline predominance of either perspective of the Necker cube, this would indicate that interhemispheric switching also mediates the alternations of this bistable perceptual phenomenon.
Finally, since the longer time course of caloric stimulation in relation to rivalry does not allow a direct assessment of the switching process itself, we used unilateral single-pulse transcranial magnetic stimulation (TMS), with its high temporal precision, to assess whether this could perturb the rivalry process. The predictions for this experiment are: (i) disruption of a hemisphereís designated percept (by TMS applied to temporo-parietal cortex) would occur only if the TMS is applied during perceptual dominance of that image; (ii) disruption of a hemisphereís designated image should have little effect on perceptual alternations if the TMS is applied when that image is perceptually suppressed (figure 2). Thus a phase-specific pattern of interference effects is expected from unilateral TMS if binocular rivalry is indeed an interhemispheric switching phenomenon.
Figure 2 Set-up for transcranial magnetic stimulation (TMS)
and binocular rivalry experiments and the perceptual interference effects
predicted by the interhemispheric switch hypothesis. (a) The circular
coil delivers a single pulse to the temporo-parietal region of the left
hemisphere. The subject views orthogonal stationary gratings (see methods
for details of the display used to avoid interaction with the intense magnetic
field) and reports their perceptual alternations using two response keys,
one of which triggers the magnetic stimulation. (b) The time course
of perceptual alternations shows the predicted disruptive effects of TMS
triggered by a switch to the horizontal percept. If the left hemisphere
adopts the horizontal percept, TMS applied to this hemisphere when the
horizontal image is perceptually dominant will disrupt this percept and
allow the vertical percept to assume dominance. The theoretical frequency
histogram (right) therefore depicts very short horizontal interval durations.
When the stimulation is delivered under identical conditions, but at the
opposite phase of the perceptual switch (i.e. triggered when the subject
reports a switch to vertical), disruption of the left hemisphere has little
effect since it is the right hemisphere that is responsible for the vertical
percept. Thus the theoretical frequency histogram (right)for this
contingency shows normal interval durations. Although not shown by theoretical
frequency histograms, it follows that if another subject shows shortened
vertical interval durations following left hemisphere TMS in one contingency,
and no effect in the other, this would indicate that the left hemisphere
has adopted the vertical rather than the horizontal percept. Actual rather
than theoretical frequency histograms are shown in figure 7.
The effect of caloric-induced left hemisphere activation on two subjectsí rivalry alternations with drifting vertical and horizontal gratings is demonstrated in figure 4 where it can be seen that the stimulation produces a change in image predominance, reflected by the V/H ratio, the ratio of total time spent perceiving the vertical and horizontal gratings, excluding mixed percepts. In individuals, the effect ranged from strong to absent (figure 5a,b), perhaps because of variation in the duration and efficacy of the procedure. The group analysis compared the absolute magnitude of change in the log-transformed V/H ratio between two pre-stimulation blocks of rivalry (a measure of the random fluctuation in V/H ratio) with the change between the block immediately prior to, and immediately following, the stimulation (a measure of the experimental effect plus random variation).
The left hemisphere activation group demonstrated a statistically significant
greater change in the V/H ratio following stimulation than was observed
in baseline viewing (figure 5a,b). This effect had largely diminished by
the fifth block of rivalry (i.e. 10-20 minutes following stimulation).
Predominance comparisons were not significant for a control group of twelve
subjects who underwent the entire protocol minus the caloric stimulation
(figure 5a). Right hemisphere activation did not induce a change in image
predominance above baseline fluctuations (figure 5a).
Figure 3 Analysis procedure for caloric stimulation experiments. There are six blocks of rivalry each representing approximately seven minutes of viewing. Each block contains four 100-second trials separated by 30-second rest periods and each block is separated by a 2-minute rest period. The first block is considered training and discarded before analysis. Blocks 2 and 3 are pre-stimulation blocks, while 4, 5 and 6 are post-stimulation blocks. The predominance ratio is calculated by dividing the total time spent perceiving the vertical gratings by the total time spent perceiving the horizontal gratings, excluding mixed percepts. Similar ratios are calculated for the oblique rivalry and Necker cube experiments. The ratios are log-transformed before analysis. There is random variation in these predominance ratios between two pre-stimulation blocks. Therefore, to show an effect of caloric stimulation, there must be greater absolute magnitude of change in the predominance ratio between blocks 3 and 4 (random variation plus experimental effect) compared with the random variation seen between blocks 2 and 3. Thus the graphs to the right in figure 5 show the ?D log predominance? for blocks 2-3 and for blocks 3-4. Subtracting the predominance changes seen between blocks 2-3 from those between blocks 3-4 removes the baseline noise and is labelled D (?D log predominance? ) in the graphs to the left in figure 5.
As a control for possible effects on image predominance from ongoing, undetected eye movements induced by the caloric stimulation, the experiments were repeated with rivalling oblique gratings. Any effect from horizontal eye movements would be spread equally across two orthogonal oblique gratings and could not therefore affect image predominance. The results of these experiments were the same as those for horizontal and vertical gratings. Left hemisphere activation significantly changed predominance above baseline fluctuations (figure 5c,d) and the effect had diminished by the fifth block of rivalry. Right hemisphere activation again did not induce a significantly greater change in predominance above baseline fluctuations, and the control condition was also non-significant (figure 5c).
To assess the direction of change in image predominance following left hemisphere activation, we looked at the twelve subjects with the largest caloric-induced shifts. For horizontal and vertical rivalry, of the 12 leftmost subjects shown in red in figure 5a (excluding the first subject who was left-handed), nine showed caloric-induced shifts towards perception of the horizontal grating while three showed shifts favouring the vertical grating. Similarly, in the oblique experiments, of the twelve subjects showing the largest predominance shifts (the leftmost subjects in red in figure 5c), nine favoured the rightward tilted (45°) grating, and three favoured the leftward tilted (-45°) orientation. Summary statistics for all caloric stimulation experiments are presented in table 1.
Figure 4 Effects of caloric vestibular stimulation on two individualsí perceptual alternations in binocular rivalry. In both cases the predominance of one perceptual alternative is shifted by left hemisphere activation (right ear cold caloric stimulation). These changes are demonstrated in the frequency histograms of interval durations for each rivalling image. The shifts are also reflected by the predominance ratios shown in the top right hand corner of each of the histograms. (a,b) This subject demonstrates a baseline horizontal predominance of V/H=0.93 which was increased to V/H=0.54 following caloric stimulation (which represents a 3-4 ?D log predominance? of 0.236). This was the usual direction of change for left hemisphere activation. (c,d) The second subject also illustrates a post-stimulation change, beginning with baseline horizontal predominance of V/H=0.94 which was reversed to a vertical predominance of V/H=1.26 (representing a 3-4 ?D log predominance? of 0.127). The direction of shift for this subject occurred in only 3 of the 12 subjects with the most marked predominance shifts, and suggests that designation of image to hemisphere is not always fixed. The effect of caloric-induced unilateral (left) hemisphere activation on the predominance of rivalling images supports the interhemispheric switch hypothesis.
In the Necker cube experiments, the effect of left hemisphere activation was dramatic in two subjects out of the 28. Each of the two subjects had normal baseline perceptual alternations, but demonstrated a virtually complete inability to see one of the two possible perspectives following caloric stimulation. One of these subjects is illustrated in figure 6c. His post-stimulation perception alternated between one clear perspective and the 'undecided'/indeterminate option where no depth was perceived in the line diagram.
Other subjects showed predominance shifts following left hemisphere
activation (e.g. figure 6a,b) similar to, and generally more pronounced
than, the effect seen with binocular rivalry. The group analysis of these
remaining 26 subjects showed that left hemisphere activation caused a significant
change in perspective predominance greater than baseline fluctuations (figure
5e,f), and that the effect had diminished by the fifth block of data collection.
Both control and sham stimulation conditions were non-significant and right
hemisphere activation did not change Necker cube perspective predominance
above baseline fluctuations (figure 5e). Of the twelve subjects with the
largest predominance shifts, seven demonstrated shifts in predominance
towards one perspective while the remaining five subjects showed shifts
in the opposite direction.
Figure 5 The effect of caloric stimulation on perceptual alternations during (a,b) binocular rivalry with drifting horizontal and vertical gratings, (c,d) binocular rivalry with stationary oblique gratings, and (e,f) viewing of the Necker cube. In all experiments, activation of the left hemisphere significantly changed baseline perceptual predominance of one image or perspective relative to the other (red inverted triangles in figures on the left). The figures to the right demonstrate the absolute magnitude of change in the log-transformed ratio of perceptual predominance (i.e. ?D log predominance? ) between blocks 2-3 (baseline random fluctuation; pink triangles) and between blocks 3-4 (experimental effect plus random fluctuation; purple triangles) for left hemisphere activation. Each point along the x-axis represents an individual subjectís data and subjects are ordered according to the magnitude of the caloric effect. There is considerable baseline noise but a majority of subjects show a shift in predominance following left hemisphere activation in all experiments (see also table 1). Effects seem to be stronger with the Necker cube than with binocular rivalry and three Necker subjects had such strong effects they could not be shown on this graph (one is described in the legend box below 5e, and for a description of the other two, see figure 6c). The figures to the left show the data for left and right hemisphere activation and the control condition that did not involve stimulation. Each point in these plots is calculated by subtracting the predominance change between blocks 2-3 from that between blocks 3-4 i.e. D (?D log predominance? ). Thus points above the zero line represent individuals who showed greater predominance change following stimulation than in baseline viewing, while points below the zero line indicate greater random change in predominance than that seen following stimulation. The subjects are arranged in descending order of magnitude and therefore, in the oblique rivalry and Necker cube experiments, the data point for an individual subject in one group does not necessarily correspond to the same subjectís data in the other two groups.
Figure 6 The effect of caloric vestibular stimulation on perceptual alternations of a reversible figure, the Necker cube. (a,b) Left hemisphere activation (right ear caloric stimulation) shifts a baseline perspective predominance (A=lower square face closer to observer; B=upper square face closer to observer; ratio calculated as for binocular rivalry and excludes indeterminate percepts) of A/B=1.3 to A/B=0.85. This represents a 3-4 ?D log predominance? of 0.185. Overall, subjects demonstrated shifts in both directions following stimulation, indicating that, unlike for binocular rivalry, designation of perceptual configuration to hemisphere may be arbitrary. Also shown (c) are the raw time series data for a single subject demonstrating the normal baseline perceptual alternations, with roughly equal time spent experiencing each perspective, followed by the effect of caloric stimulation which virtually eliminated the ability to perceive one of the two perspectives. The subject alternated between perspective A and the ëundecidedí response option (where no depth was perceived in the line diagram) following left hemisphere activation. This more dramatic effect may be related to the fact that this subject received prolonged iced water irrigation compared with other subjects. The same effect was seen in one other subject, so the effects of caloric stimulation on the predominance of Necker cube perspectives vary from infinity (2 subjects) to the more graded effects seen in the 26 subjects shown in figure 5e,f. The effect of unilateral (left) hemisphere activation on Necker cube alternations is further evidence that binocular rivalry and reversible figures have a common neural mechanism and suggests to us that this mechanism is interhemispheric switching.
Transcranial magnetic stimulation
Since left hemisphere activation showed a clear effect on both rivalry and reversible figure alternations, we concentrated on this hemisphere for the TMS experiments. Figure 7 shows that application of a TMS pulse to the temporo-parietal region of the left hemisphere had a disruptive effect on binocular rivalry which was, as predicted, phase-specific. TMS applied just as the percept was switching from vertical to horizontal caused a reversion to vertical indicated by shortened horizontal interval durations, but there was no disruptive effect when the TMS pulse was timed to occur at the opposite perceptual switch. The data for all subjects are shown in table 2 where it can be seen that this pattern occurred in three subjects. In two other subjects, TMS delivered on a switch from horizontal to vertical caused a reversion to horizontal indicated by shortened vertical interval durations, but there was no similar perceptual disruption when TMS was delivered on a switch to horizontal in these same subjects. Clear phase-specific disruptive effects of TMS thus occurred in five out of the seven subjects we tested, despite the difficulties associated with simultaneously establishing a threshold stimulation intensity and an optimal location.
Figure 7 The effect of left hemisphere transcranial magnetic stimulation on a single subject's binocular rivalry alternations (subject 1 in table 2). There is a marked disruption of percepts when left hemisphere TMS is contingent on one direction of perceptual switch but not when the contingency is at the opposite phase. (a) Control session with no TMS. (b) TMS delivered when the subject signalled a switch from the vertical to the horizontal percept, caused an immediate reversion to the vertical percept, indicated by a dramatic shortening of the horizontal interval durations. (c) TMS administered at the opposite phase (when the subject signalled a switch from the horizontal to the vertical percept), did not cause perceptual disruption. Such phase-specific disruption effects occurred in five of the seven subjects we tested (see table 2). This result supports the interhemispheric switch hypothesis and cannot be explained using a within-hemisphere competition model.
Interhemispheric switching mediates perceptual rivalry
Our results demonstrate that unilateral (left) hemisphere activation by caloric stimulation influences the alternation patterns of binocular rivalry with drifting vertical and horizontal gratings and with stationary oblique gratings. A change in the perceptual predominance of the rivalling images following unilateral hemispheric activation, is predicted by the interhemispheric switch hypothesis of binocular rivalry and is not explicable by models based on within-hemisphere competition. This interhemispheric switch hypothesis is consistent with suggestions that it is the stimulus representations rather than the eyes that rival during binocular rivalry [7-13] and that rivalry is occurring high in the visual pathway [3-17].
In further support of an interhemispheric switch model of binocular rivalry we have demonstrated a phase-specific disruptive effect of unilateral (left) transcranial magnetic stimulation on perceptual alternations. One stimulation contingency caused perceptual disruption while stimulation at the opposite phase had little effect even though delivered under identical conditions. These results cannot be explained by within-hemisphere models but are predicted by the hypothesis proposed in the present paper.
A similar effect of caloric-induced left hemisphere activation on the predominance of perceived perspectives of the Necker cube was demonstrated and supports the notion of a common neural mechanism for both binocular rivalry and reversible figures. The data presented here can be explained if both types of perceptual rivalry are mediated by an interhemispheric switch mechanism. Thus we suggest that in perceptual rivalry, each hemisphere adopts one image or perspective, and perceptual alternations reflect hemispheric alternations and thus competition between the hemispheres for visual awareness.
The lack of a change in predominance above baseline fluctuations for the right hemisphere activation group, in all three caloric stimulation experiments, may be explained in the following way. A recent fMRI study of humans undergoing binocular rivalry found a right-sided fronto-parietal activation asymmetry . This study did not separately analyze regional activation for each of the two perceptual states and hence could not assess the interhemispheric switch hypothesis. However, since activation patterns were assessed by combining both directions of perceptual switch, the finding of asymmetric activation in right fronto-parietal cortex suggests that these regions are involved in gating perceptual alternations or selecting the neuronal representations for access to visual awareness . This notion is supported by reports that right-sided frontal lesions cause the perception of only one of the two possibilities in reversible figures . The finding of right fronto-parietal activation during perceptual rivalry also emphasizes that regions involved in the gating or selection process may be functionally quite distinct from the visual regions responsible for the alternative image representations [14,17]. Left ear cold caloric stimulation might activate both the control and the visual regions in the right hemisphere and this dual activation may be responsible for the lack of predominance change in this group.
The directions of shifts in predominance induced by left hemisphere activation also raise interesting issues. There appears to be a predilection for the horizontal grating to be adopted by the left hemisphere although this was not always the case. The direction of predominance change in the oblique rivalry experiment was also biased, towards the right-tilted (45° ) orientation. It is interesting to note that both the horizontal grating and the right-tilted oblique grating were presented to the right eye. Thus eye-of-presentation may influence which hemisphere adopts which image. This is consistent with evidence that there is a higher proportion of binocular neurones with a dominant input from the contralateral eye . However, since an individualís predominance does not completely reverse in preliminary experiments in which the eye-of-presentation has been reversed, we cannot yet rule out some combination of eye-of-origin and higher-order effects. Thus the horizontal grating may often be adopted by the left hemisphere due to a cultural bias for horizontal scripts and the left-lateralization of sentence reading . The direction of predominance shifts for the Necker cube experiments suggest that with these stimuli (which do not involve separate presentation to the eyes) there may be an arbitrary designation of perspective to hemisphere. Future experiments might repeat stimulation in the same subject to elucidate whether the designation of image or perspective to hemisphere, is fixed or varies within an individual.
Table 2 Individual subject data demonstrate the phase-specific effect of TMS during binocular rivalry. The bolded median interval durations, all less than 1 second, show perceptual disruption of the interval immediately following TMS (Hs and Vs;compare these intervals with the same subjectís control intervals). Note that subjects 1-3 have perceptual disruption of the horizontal intervals when TMS is triggered by a switch to horizontal, but no perceptual disruption when TMS is triggered by a switch to vertical (see also figure 7). Subjects 4 and 5 on the other hand, experience perceptual disruption when TMS is triggered on a switch to vertical but no disruption when TMS is triggered at the opposite phase. This suggests that subjects 1-3 have the horizontal percept in their left hemisphere while subjects 4 and 5 have the vertical percept in this hemisphere. Despite this difference, subjects 1-5 demonstrate phase-specific effects of TMS and thus support the interhemispheric switch hypothesis. The statistical test compares the number of horizontal and vertical intervals less than 1-second duration in each of the two stimulation contingencies. Binomial expansions  were calculated to establish the probability of obtaining the observed distributions through chance.
Hemifields and hemispheres
In thinking about our model of interhemispheric switching, it is important not to be limited by spatially-symmetric notions of hemifield representations in V1. It has been suggested that the coherence rivalry demonstrated by Diaz-Caneja [8,9] rules out the possibility that rivalry occurs between each cerebral hemisphere . However the 1.5-degree stimulus used in the rivalry experiments reported here produces bilateral activation even in V1 (where binocular overlap is around one degree in the foveal region of higher primates) and in MT (where overlap is around 5 degrees). Moreover, the Diaz-Caneja experiment says nothing about interhemispheric competition at higher processing levels. The binocular neurones in inferotemporal cortex whose activity correlates with monkeysí reported percepts , can process information presented to either hemifield as indicated by their properties of bilateral receptive fields and ipsilateral field loss following section of the posterior corpus callosum and anterior commissure .
Thus rivalry between the hemispheres at a level beyond V1 is compatible with Diaz-Canejaís results and may actually help to explain the phenomenon of coherence rivalry. Diaz-Canejaís experiments [8,9] and the patchwork experiments of Kovacs et al. , suggest that the brain is able to group or bind coherent image segments irrespective of their eye-of-origin. How might the brain achieve such reorganization of presented image components into rivalling coherent images? The interhemispheric switch model suggests that the brain groups or binds the segments of each coherent image in separate hemispheres. Thus the perceptual resources of each hemisphere may be independently and alternately employed to achieve this kind of synthetic ability.
In the horizontal and vertical rivalry experiment, despite the care we took to delay post-stimulation testing until nystagmus had ceased, it is at least possible that the observed predominance shifts actually result from ongoing, undetected horizontal nystagmus that acts to reduce the spatial frequency and contrast of the vertical grating. The fact that three subjects had increased predominance of the vertical grating after caloric stimulation makes this explanation unlikely. Moreover, the results of the oblique rivalry experiment strongly argue against this interpretation. Results for the Necker cube experiments are also difficult to explain by eye movements. Finally, in the TMS experiments, the stimulation was delivered under exactly the same conditions for both stimulation contingencies, and any effect due to eye movements should therefore be seen in both contingencies. This was clearly not the case, as illustrated in figure 7.
Brainstem oscillator or corpus callosum?
The highly-developed corpus callosum connecting the human hemispheres may immediately suggest itself for a key role in the proposed interhemispheric switch. We think that this is unlikely and predict that split-brain subjects would still experience perceptual alternations. We suggest that the primary mechanism of interhemispheric switching involves different subcortical bistable oscillator circuits related either to the short-period perceptual alternations studied here or to long-period alternating hemispheric activity [21,22]. The suggestion that a subcortical bistable oscillator mediates interhemispheric switching is based on both comparative considerations and clinical evidence in humans.
Bistable oscillators are well-studied in invertebrates  and interhemispheric switching has been observed in the brains of birds  and fish  that lack a corpus callosum. Moreover, in human patients with midline cerebellar or brainstem damage, a roughly 90-second oscillator has been described that shows side-to-side alternation of eye movements . This oculomotor alternation, known as periodic alternating nystagmus, is believed to be a brainstem phenomenon and is accompanied by perceptual alternations during binocular rivalry consistent with our proposals concerning interhemispheric switching (Miller and Pettigrew, in preparation).
The role of the brainstem in mediating synchronous neural activity  will be particularly interesting if temporal correlation  of neurones with similar preferred stimuli is shown to be important at high levels of the visual pathway during binocular rivalry. A brainstem oscillator might increase response synchronization of neurones with similar preferred stimuli in one hemisphere, before switching its output to the opposite hemisphere to bind neurones preferring the other image. Thus simultaneous bilateral recordings from single neurones and pairs of neurones high in the visual cortex during rivalry in alert monkeys would enable testing of the interhemispheric switch hypothesis through analysis of both the rate and temporal correlation of neural activity. Other means of verifying the hypothesis include looking for the presence of alternating patterns of cerebral activation (and coherence) with electroencephalography, MEG or fMRI. It will be necessary for such studies to analyze signals derived while one percept is dominant separately from those generated during its suppression.
We have presented a readily testable neurophysiological model of binocular rivalry and reversible figure alternations based on the perceptual interference effects we have observed following unilateral hemisphere activation and disruption. Our results suggest that during perceptual rivalry, each hemisphere adopts one of the competing images or perspectives, and perceptual alternations correspond to hemispheric alternations. The interhemispheric switch hypothesis has clinical relevance because of the findings that patients with bipolar disorder (manic depression) have a slow switch rate for both binocular rivalry  and reversible figures . Our model may therefore offer a link between such findings and the emerging picture of hemispheric asymmetries in the generation and treatment of mood disorders [reviewed in 21; and see 39-41]. Finally, the hypothesis of interhemispheric switching raises new issues for the scientific study of consciousness. At any one time during perceptual rivalry, the perceived visual scene may depend on neural activity in only one hemisphereís higher visual regions.
Materials and Methods
Horizontal and vertical binocular rivalry: Eighteen right-handed and two left-handed, male and female subjects ranging from 18 to 54 years of age underwent cold caloric stimulation of the right ear (left hemisphere). Fourteen right-handed male and female subjects of similar age had left ear (right hemisphere) stimulation. Twelve control subjects underwent the full protocol minus the stimulation. Written, informed consent was obtained according to a protocol approved by the University of Queensland's Medical Research Ethics Committee. A VisionWorks™ display with liquid crystal shutters was used to present an upward-drifting horizontal (square-wave) grating to the right eye while simultaneously presenting a right-drifting vertical grating to the left eye (figure 1). The liquid crystal shutters allow the fields of view for each eye to be superimposed, with both horizontal and vertical targets occupying the same spatial location, so no training in fixation was required.
The stimuli were presented in a circular patch and subtended 1.5° of visual angle with a spatial frequency of 8c/deg moving at 4c/sec. Contrast of the gratings was 0.9. Subjects sat three metres from the monochrome computer monitor (green, P46 phosphor, persistence=500ns) and recorded their perceptual alternations by pressing one of three keyboard response buttons for vertical, horizontal or mixed percepts. The latter were removed before analysis. Baseline perceptual alternations were recorded for half an hour. This was followed by the caloric stimulation (or a rest period in the control group) and a further half-hour of rivalry data was then collected. Each half-hour session was divided into three blocks, consisting of four 100-second trials. Each trial was separated by a 30-second rest period, and each block by a two-minute rest period. The first block was considered training and was discarded before analysis.
Oblique binocular rivalry: Twenty right-handed males aged 18-25 were tested on three separate occasions. Each session involved half an hour of baseline rivalry viewing and was then followed by: (i) five minutes rest, (ii) right ear caloric stimulation and, (iii) left ear caloric stimulation. The two caloric sessions were counterbalanced. A further half-hour of rivalry data was then collected. The set-up was the same as for the horizontal and vertical rivalry experiments. A rightward tilted (45°) grating was presented to the right eye and a leftward tilted (-45°) grating to the left eye. The stimulus characteristics were otherwise the same as for the horizontal and vertical gratings except that the oblique gratings were stationary.
Necker cube: Twenty-eight right-handed males aged 18-25, underwent control sessions and left hemisphere activation by right ear caloric stimulation. Two of the left hemisphere activation subjects were unable to see one of the two possible perspectives following stimulation. Their extreme results meant that they were not included in the subsequent group analysis even though they offer striking support for the interhemispheric switch hypothesis. Sixteen subjects also underwent right hemisphere activation by left ear caloric stimulation while the remaining ten subjects underwent sham caloric stimulation with body-temperature water (and thus no vestibular stimulation). Following control sessions, the order of subsequent sessions was counterbalanced. The Necker cube was presented on a matt white surface 100cm from the subject and at eye level. The cube subtended 7.6° x 7.4° (height x width) of visual angle and had a central fixation cross (0.5° x 0.5° ). Subjects were asked to maintain gaze on the fixation point and to record their perceptual alternations using a keyboard with a response key for each of the percepts and a third option for ëundecidedí or indeterminate percepts or if their gaze strayed from the fixation point. The latter were removed before analysis. Alternations were recorded for half an hour, divided into three blocks each with three 100-second trials. Each trial was separated by a 60-second break, and each block by a 4.5-minute break. Subjects then had (i) five minutes rest (control), (ii) sham stimulation using water at body temperature, or (iii) cold caloric stimulation of the right or left ear. A further half-hour of data was then collected.
Caloric stimulation: Cold (iced) water irrigation was administered by a medical practitioner using a 50ml syringe and soft silastic tubing from a butterfly cannula. Head position was 30 degrees from horizontal bringing the lateral semicircular canal into the vertical plane; the mid-sagittal plane was vertical. The tubing was inserted into the external auditory canal until it was adjacent to the tympanum. Iced water was then instilled until the subject reported vertigo and the examiner observed nystagmus (usually following 10-30ml of iced water irrigation). Subjects demonstrated nystagmus with the brisk phase in the direction contralateral to the ear stimulated. We did not have good control over the duration or the intensity of the procedure because of the pain and nausea that the experimenter was reluctant to prolong. All subjects have been included in the results irrespective of the judged efficacy of the procedure. Post-stimulation data collection began when all visible signs of nystagmus and subjective vertigo had ceased. Sham caloric stimulation was administered by irrigation with water at body temperature.
Transcranial magnetic stimulation: Single pulse TMS was applied to the left temporo-parietal cortex using a 90mm circular coil (Magstim 200™, The Magstim Company). The centre of the coil was positioned approximately 13cm from the nasion and 12cm from the mid-sagittal line and oriented to induce current flow in a posterio-anterior direction in the temporo-parietal cortex. The coil itself was held firmly against the scalp by one of the experimenters, while the feeder cables connecting the main stimulator to the coil were supported by an overhead gantry. Magnetic stimuli were triggered when the subject signalled a perceptual switch either to the vertical percept, in one trial, or to the horizontal percept in the other. The intensity of stimulation was varied between 0.66 and 1.1 T according to the subject. The rivalry apparatus used in these experiments consisted of two 1cm (diameter) by 2cm translucent plastic tubes each with a 50d lens at the proximal end, viewing a 1mm (diameter) square wave grating (8 cycles) on translucent paper at the distal end (see figure 2). The tubes were positioned by the subject on the face-plate of a safety mask so that the gratings viewed by each eye were orthogonal in orientation and viewed at the same location.
Acknowledgements: This work was supported by the National Health
and Medical Research Council of Australia and the Australian Research Council.
We thank Andrew Tilley for helpful discussions and especially thank all
subjects for their participation.
1. Wade NJ: Early Studies of Eye Dominances. Laterality 1998, 3:97-108.
2. Blake R: A Neural Theory of Binocular Rivalry. Psychol Rev 1989, 96:145-167.
3. Leopold DA, Logothetis NK: Activity changes in early visual cortex reflect monkeysí percepts during binocular rivalry. Nature 1996, 379:549-553.
4. Sheinberg DL, Logothetis NK: The role of temporal cortical areas in perceptual organization. Proc Natl Acad Sci USA 1997, 94:3408-3413.
5. Sengpiel F: Binocular Rivalry: Ambiguities resolved. Curr Biol 1997, 7:R447-R450.
6. Logothetis NK: Single units and conscious vision. Phil Trans R Soc Lond B 1998, 353:1801-1818.
7. Kovacs I, Papathomas TV, Yang M, Feher A: When the brain changes its mind: Interocular grouping during binocular rivalry. Proc Natl Acad Sci USA 1996, 93:15508-15511.
8. Diaz-Caneja E: Sur líalternance binoculaire. Annales D'Oculistique 1928, 165:721-731.
9. Ngo TT, Miller SM, Liu GB, Pettigrew JD: Binocular rivalry and perceptual coherence. Curr Biol (in press).
10. Logothetis NK, Leopold DA, Sheinberg DL: What is rivalling during binocular rivalry? Nature 1996, 380:621-624.
11. Andrews TJ, Purves D: Similarities in normal and binocularly rivalrous viewing. Proc Natl Acad Sci USA 1997, 94:9905-9908.
12. Campbell FW, Gilinsky AS, Howell ER, Riggs LA, Atkinson J: The dependence of monocular rivalry on orientation. Perception 1973, 2:123-125.
13. Atkinson J, Campbell FW, Fiorentini A, Maffei L: The dependence of monocular rivalry on spatial frequency. Perception 1973, 2:127-133.
14. Lumer ED, Friston KJ, Rees G: Neural Correlates of Perceptual Rivalry in the Human Brain. Science 1998, 280:1930-1934.
15. Tong F, Nakayama K, Vaughan JT, Kanwisher N: Binocular Rivalry and Visual Awareness in Human Extrastriate Cortex. Neuron 1998, 21:753-759.
16. Srinivasan R, Russel DP, Edelman GM, Tononi G: Increased Synchronization of Neuromagnetic Responses during Conscious Perception. J Neurosci 1999, 19:5435-5448.
17. Leopold DA, Logothetis NK: Multistable phenomena: changing views in perception. Trends in Cognitive Sciences 1999, 3:254-264.
18. Zaidel E, Clarke JM, Suyenobu B: Hemispheric independence: a paradigm case for cognitive neuroscience. In Neurobiology of Higher Cognitive Function 1990, eds. Scheibel AB, Wechsler A: The Guilford Press, New York, 297-355.
19. Luck SJ, Hillyard SA, Mangun GR, Gazzaniga MS: Independent hemispheric attentional systems mediate visual search in split-brain patients.Nature 1989, 342:543-545.
20. Ramachandran VS: Phantom Limbs, Neglect Syndromes, Repressed Memories, and Freudian Psychology. Int Rev Neurobiol 1994, 37:291-333.
21. Pettigrew JD, Miller SM: A ëstickyí interhemispheric switch in bipolar disorder? Proc R Soc Lond B 1998, 265:2141-2148.
22. Shannahoff-Khalsa D: The Ultradian Rhythm of Alternating Cerebral Hemispheric Activity. Int J Neurosci 1993, 70:285-298.
23. Suthers RA: Peripheral Control and Lateralization of Birdsong.J Neurobiol 1997, 33:632-652.
24. Pettigrew JD, Collin SP, Ott M: Convergence of highly-specialized behaviour, eye movements and visual optics in the sandlance (Teleostei) and the chameleon (Reptilia). Curr Biol 1999, 9:421-424.
25. Baloh RW, Honrubia V, Konrad HR: Periodic Alternating Nystagmus.Brain 1976, 99:11-26.
26. Bottini G, Sterzi R, Paulesu E, Vallar G, Cappa SF, Erminio F et al.: Identification of the central vestibular projections in man: a positron emission tomography activation study. Exp Brain Res 1994, 99:164-169.
27. Vitte E, Derosier C, Caritu Y, Berthoz A, Hasboun D, Soulie D: Activation of the hippocampal formation by vestibular stimulation: a functional magnetic resonance imaging study. Exp Brain Res 1996, 112:523-526.
28. Posner MI, Petersen SE: The Attention System of the Human Brain.Annu Rev Neurosci 1990, 13:25-42.
29. Vallar G, Bottini G, Rusconi ML, Sterzi R: Exploring somatosensory hemineglect by vestibular stimulation. Brain 1993, 116:71-86.
30. Walker P: Stochastic properties of binocular rivalry alternations.Percept Psychophys 1975, 18:467-473.
31. Meenan JP, Miller LA: Perceptual Flexibility after Frontal or Temporal Lobectomy. Neuropsychologia 1994, 32:1145-1149.
32. LeVay S, Connolly M, Houde J, Van Essen, DC: The complete pattern of ocular dominance stripes in the striate cortex and visual field of the macaque monkey. J Neurosci 1985, 5:486-501.
33. Bavelier D, Corina D, Jezzard P, Padmanabhan S, Clark VP, Karni A, et al.: Sentence reading: a functional MRI study at 4 Tesla.J Cogn Neurosci 1997, 9:664-686.
34. Gross CG, Rodman HR, Gochin PM, Colombo MW: Inferior Temporal Cortex as a Pattern Recognition Device. In Computational Learning and Cognition: Proceedings of the 3rd NEC Research Symposium 1993, ed. Baum E, 44-73.
35. Marder E, Calabrese RL: Principles of Rhythmic Motor Pattern Generation. Physiol Rev 1996, 76:687-717.
36. Munk MHJ, Roelfsema PR, Konig P, Engel AK, Singer W: Role of reticular activation in the modulation of intracortical synchronization.Science 1996, 272:271-274.
37. Engel AK, Roelfsema PR, Fries P, Brecht M, Singer W: Role of the Temporal Domain for Response Selection and Perceptual Binding.Cereb Cortex 1997, 7:571-582.
38. Hunt J, Guilford JP: Fluctuation of an ambiguous figure in dementia praecox and in manic-depressive patients. J Abnorm Soc Psychol 1933, 27:443-452.
39. Bejjani B-P, Damier P, Arnulf I, Thivard L, Bonnet A-M, Dormont D et al.: Transient acute depression induced by high-frequency deep-brain stimulation. New Engl J Med 1999, 340:1476-1480.
40. Pascaul-Leone A, Rubio B, Pallardo F, Catala MD: Beneficial effect of rapid-rate transcranial magnetic stimulation of left dorsolateral prefrontal cortex in drug-resistant depression. The Lancet 1996, 348:233-237.
41. Grisaru N, Chudakov B, Yaroslavsky Y, Belmaker RH: Transcranial Magnetic Stimulation in Mania: A Controlled Study. Am J Psychiatry 1998, 155:1608-1610.
42. Ferguson GA: Statistical Analysis in Psychology and Education, 3rd edn. New York: McGraw-Hill; 1971.