MIT Press (2012)
Summary/review of this book
Perception is argued to take time to build, and does not come to us all at once. The observer has to integrate bits and pieces of information dispersed across time and space. Eye movements are devoted to acquiring information about the environment, but these movements are neither random nor properly systematic. The retina is technically part of the brain, but does not receive feedback from other brain regions. Vision derives initially from head movements and eye movements. At any one moment, we can see only a small fraction of our field of vision in precise detail, for instance only the right eye of the portrait of Mona Lisa when looking at this portrait. The small area, about 1% of the visual field, seen in sharp detail, corresponds to the centre or fovea of the eye. Detailed information about the environment is obtained by continously refocusing the eye onto different locations.
The retina consists of several layers of cells at the back of the eye including photoreceptor cells. These transduce light into potentials along neuron membranes. There are two types of photoreceptor cells, rods and cones. Rods respond to low levels of light but not to colour, while cones have the opposite characteristics. The fovea at the centre of the retina is formed almost entirely of cones. The rods become more frequent the greater the distance from the fovea. The fovea provides colour and sharp vision for a small proportion of the visual field, while the peripheral rods provide less detailed vision for the majority of the visual field. Spatial orientation can take place independently of the direct orientation of the eyes and the head. This form of viewing is referred to as covert attention. Covert processing refers to neural processing or selections that are independent of the position of the head and eyes. Visual scanning of the environment is for the most part biased viewing. We pay more attention to some aspects than to others, and our gaze may return to these. In processing faces we concentrate on the eyes, mouth and the tip of the nose. Gaze is also biased towards oddities in the scene.
Dopamine neurons are involved in the response to novel or reward-related stimuli. These neurons are located in two areas of the midbrain. Those in the ventral tegmentum area project to the ventral striatum and the orbitofrontal cortex. Those in the substantia nigra project to the dorsal striatum and the dorsolateral prefrontal. The neuromodulator, dopamine, is released via these projections. This is thought to be related to both novelty and reward/punishment prediction. The ventral dopamine is activated by reward, but inhibited by punishment, but the dorsal dopamine was activated by both reward and punishment. The ventral route which distinguishes between reward and punishment is related to the brain’s emotional processing, preference formation and decision making. The route to the dorsolateral is suggested to signal overall importance rather than preference, and this area is seen as important for the orientation of gaze.
Other studies monitoring blood flow in the brain indicate a similar split of activity, with value-based decisions leading to increased flow in the orbitofrontal-related areas, while exploration for new information leads to increased flow in the dorsolateral related areas. Inhibitory output from the basal ganglia is also suggested to suppress alternatives to chosen actions, but to not inhibit explorative actions. Dopamine activity is also stronger for indications of reliable as opposed to random information. Activity is related to the reliability of information rather than the size of the expected reward, so information can be a reward in itself.
The author discusses excitatory pathways that initiate eye movements, mechanisms that prevent eye movements and also those that report back on eye movements. The superior colliculus is seen as controlling saccadic eye movements, where the gaze jumps from one point to another. Excitatory projections to the superior colliculus come from the parietal lobe and the frontal eye field (FEF). These are active in voluntary eye movements. However, there is also an inhibitory input of the neurotransmitter, GABA, via the basal ganglia. This makes the superior colliculus more resistant to excitatory inputs.
Information from the retina reaches the cortex via the lateral geniculate nucleus. The information of the rods and cones is processed separately. The parvocellular layers receive information from the cones related to colour and fine detail. The magnocellular layers refer to less detailed/ more global information even when the signal is not particularly clear. Biases can be activated by the contexts of particular situations registered in our past knowledge before any perception arises. Given contexts can activate biases, and thus guide perception.
Thus activity in the fusiform face area is greater when the face information is related to the rest of the body rather than viewed in isolation. Some of this bias processing is unconscious with subjects starting to predict regular appearances of data without having consciously realised the existence of a regularity. It has been suggested that this type of regularity processing could occur in the hippocampus and the medial temporal lobe. The context and related bias is seen as a factor in guiding the visual gaze. The firing of neurons in the inferior temporal cortex shows a bias or selectivity towards particular objects that overrides the difficulty of dealing with different views or partial occlusions of the object. Thus there is a bias towards accepting multiple views of the same object. The projection of the inferior temporal region has the best correlation with stimuli coming from objects in the external environment.
The retina does not receive feedback signals from the brain, and is thus autonomous in terms of processing the initial incoming visual information. The retina acts as an initial filter to select those signals that are more important. Signals from retinal ganglion cells are suggested to have already begun the process of separating objects from their background. Some retinal ganglion cells respond to objects that move in a different direction to their background. This may help to deal with sudden appearance and disappearances. Some retinal processing responds when an element of a regular pattern is missing.
The author touches on the question of change blindness, which was such a favourite of 1990s consciousness studies, pointing out that when this ‘blindness’ occurs, the motions that originally led to the change is being masked. Whenever they are not masked, these changes direct our gaze. It is suggested that it was adaptive to be aware of changes that occured because of current motions, rather than changes that occured out of sight. So there was not much adaptive benefit from being less subject to change blindness. The brain is above all geared to be aware of the retina signalling that an object is moving. Such new information requires inspection because ignoring it could be dangerous.
The gaze shifts faster towards faces than other objects. This involves processing in the fusiform face area. The bias towards faces and some fear objects such as snakes is viewed as being genetic, while other biases derive from the life experience. The fusiform face area can also be used for some very specialised recognition such as that of expert bird watchers. New synapses grow so as to increase biases towards particular information; thus the bias is built in before conscious processing occurs, and this speeds up the response to incoming information.
Studies have thrown interesting light on the processing of the higher visual cortex. Some neurons in the medial temporal lobe fire both for a particular image and for the imagined or remembered version of the image, indicating that these two functions use the same neural process. In another study, where subjects were presented with hybrid images, it was shown that neurons in the medial temporal lobe determined the dominance of particulat patterns in the image. But this could be influenced by the decision of subjects to focus on particular patterns, so that the subject’s decision appeared to influence the activity of the neurons. Some studies suggest that the brain projects a virtual image of future behaviour. Thus a rat at a junction in a maze has processing in the CA3 area of the hippocampus and the ventral striatum which appears to relate to prospective activity.