Brain activation and the locus of visual awareness

Abstract

A major problem in visual neuroscience is to distinguish neuronal activity which is directly related to the conscious percept. The word “directly” is used here as opposed to an indirect relationship, as is for example the case with activity in the retina, which is produced by a stimulus in the outside world and will eventually lead to the perception of this stimulus. As for the word “related”, it is used to mean activity which creates the perceptual experience or, even more extremely, activity that is the perceptual experience. The distinction between the two (is vs. creates) is not straightforward, although there might be some differences between them. Philosophers would argue that they have a different phenomenology, the percept existing only for the perceiving person, whereas the underlying neuronal activation exists for all to observe. One could go on to argue that it is actually not the neuronal activation that “everybody” observes, but each one observes his own percept of it, which is also unique and subjective. Still, the content of this percept and the one of the original stimulus are quite different. The purpose of the present review is not to dig deep into such philosophical issues, but rather to give an overview of neuroscientific approaches trying to locate the neural correlate of conscious perception.

The problem can be stated as follows: light entering the eye creates a cascade of neuronal events throughout the visual pathway, and at some point in this parallel, hierarchical and recurrent network, a conscious perceptual experience arises—where does this happen? A first, intuitive solution would be the possibility that visual consciousness arises when the neuronal signals reach some higher stage of the visual system. The serial, hierarchical view of the visual system as a system of increasing complexity, originally proposed by Hubel and Wiesel,1 points towards this end: visual consciousness is the result of some complex type of processing, where each successive stage analyzes the stimulus but at a more complex level than the antecedent one leading to a conscious experience somewhere high-up in the visual system. An influential theoretical paper supporting this view was published by Crick & Koch in 1995,2 claiming that activation of area V1 (which is the first cortical stage of visual processing) does not reach awareness. Some experimental work soon came to further support this view, showing that existence of stimulus-specific activation in V1 in the absence of a conscious experience of the stimulus. From the side of psychophysics, for example, it has been shown that the, orientation-specific, elevation of the contrast detection threshold of sinusoidal grating is still present after adaptation to a grating which is rendered invisible due to crowding.3 From the side of physiology, it has also been shown that color-selective V1 neurons can signal the alternation between red and green even at frequencies where the two colors fuse perceptually producing a constant perception of yellow.4 In the human, fMRI studies have shown that one can still get orientation-specific responses from V1, despite the use of masking that renders this orientation invisible to the participants which are being scanned.5

Perhaps the first attempt to dissociate brain activation directly related to the percept from activation related to merely processing the visual stimulus, comes from the studies of Nikos Logothetis. He had the ingenious idea of using binocular rivalry (BR) in which a constant stimulus (two different images, one in each eye) gives rise to a varying perception. By using electrophysiology to monitor neuronal activation under such circumstances, one could hope to distinguish neurons that signal the alternations from one image to the other in the perceptual experience of the subject.6–8 Result of this series of studies can be summarized as follows: (1) there are “perceptual” neurons at all the stages of the visual system, (2) the percentage of such neurons seems to increase as one moves to the higher areas of the system. For some puzzling reason, the scientific community has paid less attention to (1) than to (2), taking the BR results as evidence that visual consciousness must be a function of the higher visual areas. It must be noted here that BR has turned out to be slightly problematic as a method to study visual consciousness, because we are not yet certain about the mechanism underlying this phenomenon. It is quite likely that one of the two eyes/images/signals is blocked at some stage in the system, and thus for areas above this point it is not really true that the stimulus is “constant”—removing the stimulus from the screen would be equivalent, as far as these areas are concerned. However, as we will see, additional methods have been invented to achieve the very much desired result of dissociating the physical stimulus from the subjective percept.

The notion that the content of visual consciousness is actually reflected in the activation of areas of the visual system beyond V1 has been supported by several imaging studies in the human. In the very first of them, it was shown that a stationary stimulus that induces illusory motion to an observer is able to activate area V5 of the human brain, an area long known to be related to the processing and perception of motion.9 The same area has also been shown to be activated by the motion-after effect, where a stationary stimulus appears to move due to prior adaptation to a moving stimulus.11 (It must be said, however, that the results of this particular study were probably due to attentional differences, as shown by further experiments.10) Furthermore, the response of area V5 to a constant stimulus which is sometimes perceived as flickering and sometimes induces the perception of apparent motion, varies in accordance with these perceptual alternations.12 We have used BR together with fMRI to show that perceptual alternations between coherent motion and motion noise random dot stereograms are faithfully reflected in the activation of this area in the human.13 Finally, electrophysiological studies in the monkey have shown that, for a constant motion stimulus, responses of individual V5 neurons correlate well with the directional perception of the animal.14

As it should be clear by now, the idea that visual perception is located in the higher areas of the visual system is supported by studies showing that (a) V1 activation can be dissociated from awareness and (b) activation of these areas correlates with the percept. An obvious objection to this idea would come from studies showing that activity in V1 can is some cases very well correlate with the percept; several such studies have been done. Using human fMRI, it has been shown that fluctuation in V1 activation faithfully reflect perceptual alternations during binocular rivalry between a low and a high contrast grating.15 In a similar study, one of the two rivalling stimuli was placed in the blind spot of one eye and it was shown that the corresponding (monocularly driven) part of V1 would modulate its activity depending on whether the stimulus activating it was perceptually suppressed or not.16 In an fMRI study using binocular rivalry between two different orientations, it has been shown that the information of which one of the two orientations is perceived at each point in time can be ‘read’ in the activation pattern of human V1.17 The results from these studies collectively show that, in many cases in which the stimulus is dissociated from the percept, V1 activation reflects the latter rather than the former. There is also further evidence that area V1 might play a direct, exclusive role in consciousness: some of our perceptual capabilities, as for example sensitivity to changes in contrast or vernier acuity, can only be found in the responses of V1 neurons and are totally lost [Unless they are coded in some other, more complicated way than the firing rate of single neurons (e.g., correlation patterns of neuronal populations)] in the neuronal responses of higher areas.18 Furthermore, the strength of adaptation aftereffects that, at least partly, seem to relate to neural events in V1, is modulated by the amount of time the adapting stimulus was perceived.19

A common argument against the V1 fMRI results presented above is that the reported perceptual correlations of neuronal activation could be a result of feedback from higher areas, where consciousness is created, or so its adherents claim. Therefore, in order to further show that the link between activation of higher areas and perception is not so strong, one needs to show that activity in higher visual areas can also be dissociated from the perceptual experience. In one such study, we have investigated into the relationship between the activation of the Fusiform Face Area (FFA) and Parahipocampal Place Area (PPA) and the perception of faces and houses. These areas have been previously shown to be selectively activated by face and house stimuli respectively,20 and also to reflect the content of the alternating percept during binocular rivalry.21 We have used dichoptic color fusion to make such stimuli invisible to subjects inside an fMRI scanner. Despite the fact that subjects could not distinguish between the faces and houses presented to them, areas FFA and PPA showed stimulus-specific activation elicited by these invisible stimuli.22 The methodology is the opposite of what is done with binocular rivalry, where a constant stimulus gives rise to a varying perception. Instead, perception is kept constant (a uniform yellow field) whereas stimuli change between faces and houses. Despite these changes not being consciously perceived, they modulate activation in these higher visual areas. We have also discovered a similar effect in area V5, which is one of the most extensively studied areas in the context of visual perception with activity in it commonly reflecting the content of the visual experience (see above). This area is activated more from a moving than from a flickering grating. By using crowding to render the motion of such a peripherally-presented grating inaccessible to consciousness, we have shown that V5 still modulates its activity in response to changes in the visual stimulus that do not reach awareness.23 Therefore, stimulus-specific activation of higher visual areas might be necessary but does not seem to be sufficient for visual awareness.

What is then the role of visual areas beyond V1 with respect to visual processing and perception? As already described in the previous paragraphs, their activation can be modulated by perceptual changes which are not due to a change in the visual stimulus, but also by changes in the stimulus which do not reach perceptual awareness. Therefore, both the stimulus and the percept seem to be able to influence the response of these areas in an independent way. But which one of the two is the most dominant in dictating the pattern of an areas response? In a paper published recently in PNAS, we have tried to answer this question with respect to area V5 which is, as already mentioned, the most well studied area in relation to visual motion.24 We have used one stimulus which consisted of presenting two strong monocular motions of opposite directions, one in each eye. The two monocular stimuli where constructed in such a way that, instead of producing binocular rivalry, they cancelled each other out at the binocular level. This stimulus was therefore characterized by strong motion energy at the physical level, but a very weak, almost absent sensation of motion at the perceptual level. We used fMRI to compare V5 activation produced by this stimulus to a normal motion stimulus in which the same directional motion was presented to both eyes, resulting in a very clear motion percept. In other words, we used a condition in which a very strong motion stimulus resulted in a very weak motion percept and a condition in which a less strong motion stimulus resulted in a much stronger motion percept. By pitting stimulus vs. perception in this way, we discovered that it is the former, not the latter that dictates the response of V5—the area gave a stronger response in the case of the weaker percept. Therefore, this area not only seems to be involved in both the sensory and the perceptual representation of the motion, by is also biased more towards the former and less towards the latter. This human fMRI result is in agreement with electrophysiological experiments in the monkey, in which the response of V5 neurons to the local motion of elements of the stimulus was compared to the global motion of the whole stimulus: the two motions had opposite directions and, despite perception being dominated by the direction of the global stimulus, neurons were found to signal the direction of the local elements.25 These results collectively call for a re-evaluation of the role of higher visual areas, in the perception of the visual attributes that they are functionally-specialized to process.