Visual Processing in the Human Cerebral Cortex:
Neuromagnetic Studies


Summary and conclusions


Visually evoked responses and reactivity of spontaneous brain rhythms to visual stimuli were investigated in human observers with a Neurornag-122 whole-scalp neuromagnetometer. Behavioral responses were measured during MEG recordings in three out of five studies. The work was performed in the Brain Research Unit of the Low Temperature Laboratory, Helsinki University of Technology.

Evoked responses after meaningful visual objects and meaningless nonobjects were recorded during an object detection task. Although many brain areas showed more activation to objects than to non- objects, only the activation strength of the right lateral occipital cortex correlated with the proportion of correct object detections. This result shows that, while several brain areas participate in object processing, response selection may be specifically associated with only one area. Evaluation of the meaningfullness of the stimulus was evidently not possible without awareness of the object shape. Therefore, the results suggest that the right lateral occipital activity correlates with visual awareness of objects.

During the object detection task, the parieto-occipital alpha rhythm was stronger after non-objects than objects. Such a difference was, however, not detected during a control study, when no discrirnination was requested. The reactive alpha rhythm seemed to originate in the medial parieto- occipital sulcus. The corresponding region in monkeys processes the real position of visual objects for motor behavior and is one of the pathways through which visual information reaches parietal cortices. The result suggests that information about visual object form may affect activity in the dorsal visual stream, and that the higher level of alpha rhythm after non-objects than objects may reflect reduced transfer of information through parieto-occipital region after non-objects. A hypothesis is put forward that this phenomenon is associated with reduced sensitivity of the ventral visual stream when it is processing target identity. During this process the dorsal stream searches new targets to attend to, and may, for example, plan subsequent saccades or reaching movements.

Increase of the sensorimotor l - Hz mu rhythm level was observed after timelocked binocular rivalry. Reactivity of the mu rhythm seemed to be specific for the time-locked shift of dominance between the two eyes, which was accompanied with a change in visual percept. Similar mu rhythm enhancement was found after real changes in the stimulus, when the rivalry and real changes were presented in random order. No systematic correlation was found between mu and alpha rhythm levels. The results show that a visual stimulus can modulate activity of the primary somatosensory cortex, and suggests that this modulation follows changes in visual percepts. Functionally, such an interaction might be important for visuomotor integration, such as guiding eye or hand movements, or visuohaptic interaction serving perceptual purposes.

Despite normal intelligence and education, about every twentieth child fails to acquire adequate reading skills. The dyslexics have, according to one hypothesis, a defective magnocellular pathway which leads to defective processing of fast and transient visual information. Because visual motion processing is especially dependent on the magnocellular pathway, it would be logical to find defects in this process. While earlier behavioral findings showed subtle defects in visual motion detection in dyslexics, a recent fMRI study suggested a complete inactivity of the visual motion-processing area V5. We studied this area with three different visual motion stimuli in normal controls and dyslexic adults, and found no differences in source locations or amplitudes between the two groups. However, the latencies of the peak responses seemed somewhat longer in dyslexics. Our results contradict the earlier fMRI findings and show that in dyslexics at least some visual motion stimuli elicit rather normal activation in the visual motion-processing area.

In visual search, the observer is looking for a target pattern among simultaneously presented distractors. If the target and the distractors differ in some basic feature, such as orientation or local motion, the target "pops out" from the distractors. Evoked responses and reaction times were recorded while the observers were viewing a display in which the target was defined by an abrupt change of line orientation. The observers performed four different tasks; passive viewing, detection of any motion, locating the motion, and identifying the direction of the motion. Activity of two bilateral source regions was analyzed more closely: the occipito-temporal area, corresponding to the known location of the human visual-motion processing area V5, and the parietotemporal area. In line with earlier studies showing sensitivity of the visual motion processing area to attention, the right occipitotemporal area showed increasing activity with increasing task difficulty. No differences were found in brain activation between target localization and identification tasks. This result is interesting in light of recent behavioral models which propose that information about location and identity of a target might not be available simultaneously; while not disproving these models, our result shows that cortical processing is similar during both tasks. During the localization and identification tasks, the right hand reaction times were about 30 ms faster after right than left visual field stimuli. This shift possibly corresponds to information transfer from one hemisphere to the other. When detecting a motion with no request to identify or localize the target, the reaction times were faster, and similar for stimuli in both left and right visual hemifields. On the other hand, the latencies of the activation in the occipitotemporal area seemed to be similar for all conditions, on average longer for stimuli in the ipsilateral visual field. Thus different routes of information can be utilized for response selection during the different tasks.

In summary, these studies provided new information about visual object processing, demonstrated similar visual motion processing in normal and dyslexic readers, revealed cooperation between occipitotemporal and occipitoparietal visual streams, and documented reactivity of sensorimotor mu rhythm after a visual stimulus. In addition, in contrast to recent behavioral models cortical processing during different pop-out tasks showed similar brain activation.