Table 11.1 Comparison of language errors and visual illusions Language errors Visual illusions
Speculations
We have hardly said anything about how emotions may affect seeing, except for the empathy theory of distortions, which we dismissed in favour of very different ideas (page 208). Perhaps, now, we should listen to Lady Macbeth:
Is this a dagger which I see before me,
The handle toward my hand? Come, let me clutch thee:
I have thee not, and yet I see thee still.
Art thou not, fatal vision, sensible
To feeling as to sight? or art thou but
A dagger of the mind, a false creation.
Proceeding from the heat-oppressed brain?
Mine eyes are made the fools o’ the other senses,
Or else worth all the rest: I see thee still
And on the blade and dudgeon gouts of blood,
Which was not so before. There’s no such thing:
It is the bloody business which informs
Thus to mine eyes.
Emotions can affect perception, though not systematically (as in the illusions discussed in the previous chapter). Conversely, perceptions can affect emotion—hence powers of art. Some things look beautiful, others ugly. But after all the theories that have been put forward we have little idea why this should be so. The answer must lie far back in the history of human experience—no doubt from pre-human dramas of successes and failures of survival. For presumably we find flowers and sunny days beautiful because, through millions of years, they enhanced living and increased chances of creating new life. So we resonate with flower-seeking insects, but hardly with creatures of the dark. Our appreciation of symbols has immeasurably extended ancient responses; so surrogate pictures and words and music evoke ancient emotions related to surviving and reproducing ourselves.
It has been suggested that emotions are sensations of bodily adaptations to situatioins of danger, love, or whatever. This is known as the James–Lange theory, from the American psychologist William James (1842–1910) and Swedish physiologist C. G. Lange (1834–1900). They suggested that emotions do not cause bodily changes, but rather the other way round. For example a flood of adrenaline for the ‘flight or fight’ reaction to a dangerous situation causes the sensations of fear. This means that feelings of emotions are perceptions of reactions of our bodies, to situations they responded to aeons ago. Although this may not explain all the subtleties of emotions, it does seem to be their biological basis.
In this account, emotions are intimate perceptions of bodily changes; but their significance is very much in terms of social situations and how they are perceived. Charles Darwin suggested that the red cheeks and neck of blushing are a public signal that the individual has transgressed a social rule and so is not to be trusted. The red flush gives the game away, while the accompanying sense of confusion makes it hard to concoct an excuse. Blushing is unique to humans and does not occur in childhood before social mores for ‘good’ and ‘bad’ behaviour are appreciated.
Whatever their ancient origins, emotions are deeply associated with meaning. A recurring theme of this book is the importance of meaning for perception—and also of sophisticated perception for perceiving meanings. This develops from relatively meaningless patterns to seeing potentialities of objects. Object-meaning transfers to pictures, though objects seen in pictures are powerless to hurt or to reward, and we are but passive spectators of pictured worlds. Objects are defined initially by what they can do to us and we to them. Almost whatever a table looks like it is an object for putting things on, and a chair is for sitting on. Certain expectations must be fulfilled. If a book were placed on a supposed table which melted away, we would say that it was not after all a table, or any object. It might be written off as a dream or hallucination. Here mirror images are interesting, for though objects are seen, they are dislocated in space and divorced from behaviour. All pictures of objects are transposed in space and time.
Although the sensory dimensions of sight, touch, smell, and so on, are very different, we have no hesitation in accepting that they all belong to the same world of objects. And clearly our knowledge of objects is not limited to sensory experience. We know about magnetism though we cannot sense it, and about atoms though they are invisible. An essential power of science is to extend knowledge far beyond sensory perception—often to change how we see.
It is not hard to guess why the Intelligent Eye creates hypotheses beyond visual information; as behaviour can be directed not only to what is sensed but also to what is hidden, and to what may happen in the immediate future. The richness of perceptual hypotheses (perceptions) confers immense survival benefits, as well as making the world and illusions perceptually and conceptually interesting.
If the brain were unable to fill in gaps, and to bet on limited evidence, behaviour would come to a halt in the absence of directly relevant data from the senses. The cost of going beyond the evidence—to see and live by predictive hypotheses—is a rich variety of illusions generated by intelligence. Intelligence is dangerous, but stupidity is worse! Lack of intelligence is worse because it limits possibilities. We might say that perceptual intelligence modifies the computer adage ‘Garbage in—garbage out’, to ‘Garbage in—sense out’. But evidently this is far from infallible, for many illusions and errors are generated by seeking sense from sensation.
Let’s try to bring our too-complicated discussions of phenomenal phenomena of illusions to a focus—by attempting to classify them. Why should we go to this further trouble? Classifying is extremely important for any science. Mendeleyev’s periodic table of the elements made sense of chemistry, its gaps suggesting where to look for new advances. Classifying species in biology, and kinds of stars in astronomy, led to understanding their evolution. Classifications can separate useful from misleading analogies. They structure present knowledge and they guide new observation and experiment.
Even more basic: at least implicit classes are needed for building up generalized knowledge from instances by induction. Explicitly defined classes are also essential for deduction—for inferring particular conclusions from generalizations, with formal rules.
The simplest animal learning is inductive, including Pavlovian conditioning—such as relating a buzzer to food. No doubt this is why much of learning is slow: relevant associations have to be discovered through many instances. Rules of deduction are relatively recent: formulated by Aristotle for his syllogisms.
Possibly this attempt to classify phenomenal phenomena of vision will help to structure the unnatural science of illusion. Its gaps and limitations may suggest observations and experiments, for further understanding of how we see. So, we will end by:
Putting illusions in their place
Classifications change with shifts of theoretical understanding or opinion. An almost too obvious biological example of this is the whale. Looking like a huge fish, it came to be classified as a mammal, as it is air breathing, bears its young, and so on. Initial appearances give way to needs of classification. More detailed classes are developed when more discriminations are needed, generating technical terms which, though useful for experts can be barriers to general communication. This is especially so when the criteria for belonging to classes are not simple appearances. Here, the exercise of trying to classify should make our understanding more explicit, and it may suggest ‘litmus tests’ for deciding where to put phenomena of perception in their place.
The appearances fall neatly into four classes, which we have called: ambiguities, distortions, paradoxes, fictions. It may be noted that these are also errors of language. This may be no accident. Perhaps human language developed from pre-human perceptual classifications of objects and actions. This notion might explain why the natural languages have similar basic structures. This is stressed by experts such as Noam Chomsky and Steven Pinker, though such innateness is perhaps controversial.
We start with the physical world: the source of signals to the senses. The first source of visual illusions lies in disturbances of light before the eye is reached. These we may call physical illusions. Though whether there is an illusion depends on whether or not the perceiver corrects for the optical disturbance.
Sensory signals are carried by a great variety of parallel neural channels—for contours, movement, colours, stereo depth, and so on—and are processed by specialized brain modules (Chapter 4). Upsets of the physiological channels or the brain processing modules may generate physiological illusions.
The signals are ‘read’ in two ways, to give useful data: through general rules, and by knowledge of objects. Using rules and knowledge can be called ‘cognitive.’ When rules or knowledge are misapplied, illusions occur through no ‘fault’ of the physiology. So although physiology is always involved, it is not always responsible for errors or illusions. It is what the physiology is doing—and whether this is appropriate—that matters.
We may call the signals from the senses ‘bottom-up’, and the application of knowledge (which may or may not be appropriate) ‘top-down’. Here I shall add ‘sideways’ for rules. (This by rough analogy with putting floppy disk programs ‘sideways’ into a computer; though we are not assuming the brain to be a digital computer. It seems, rather, to be analogue modules of neural nets.)
We have hinted that perhaps the essential structure of languages derived from ancient, pre-human, perceptual classifications of objects and actions. For it is striking that the obvious names for kinds of illusions are the same as for errors of language: ambiguities, distortions, paradoxes, fictions (Table 11.1).
Table 11.1 Comparison of language errors and visual illusions Language errors Visual illusions
We have found also four basic causal classes of illusions: physical, physiological, knowledge, rules. Let’s construct a classification—in terms of appearances and kinds of causes—which might be helpful for understanding, and for suggesting new observations and experiments.
Physical optical illusions have their cause before the eye is reached. They do not tell us anything about physiology or cognition; except, indeed, that they are not countered by cognitive visual processes, or checks, or by understanding of the situation. The lack of checks is presumably because perception has to work extremely fast to be useful in real time. (Knowledge, for example of reflections, does not counter mirror illusions.)
In general, we might say of phenomenal phenomena that physiological illusions tell us about brain, and cognitive illusions about mind. (This, if we think of mind as the brain’s knowledge and hypothesis-generating rules and intelligence.)
Classes of illusions
Table 11.2 shows a tentative classification in these terms of phenomenal phenomena of vision. It represents tentative decisions for putting these phenomena in their place. How can we decide more objectively? We need more explicit ‘litmus tests’ for distinguishing and classifying phenomenal phenomena. I hope to develop this in a forthcoming, more technical book on illusions.
If we put these various ideas together, we arrive at a scheme something like Figure 11.1. (This is a development of Figure 10.9.)
Perceptual brain systems do not always agree with the thinking cortex. Thus, for the cortex educated by physics, the moon is about 400 000 km distant (nearly a quarter of a million miles), but to the visual brain it appears but a few hundred metres away. Although we know the distance intellectually, the visual brain is never informed, so we continue to see the moon as almost within our grasp. This deep separation between conceptions and perceptions is disturbing, yet no doubt inspires much of science and art.
In spite of many experiments—including Stratton’s experience with inverting goggles (page 141), Held and Hein’s kittens in baskets (page 144), and adult recovery from infant blindness (pages 151–158)—we still know little of how perceptual learning works; yet relations between perceptual and conceptual learning are obviously important for education. Interacting—playing—with objects develops perceiving and conceiving. So, hands-on science centres may help children and extend perceptual and conceptual learning through adult life. Figure 11.1 shows feedback from handling objects to develop perceptual and conceptual knowledge. Much of our normal hands-on experience from infancy onwards is, however, misleading for understanding physics (tending to make us think in Aristotelian terms) because, for example, toy cars (and real cars) need a continuous force to keep them going. This, because friction contaminates and hides the deep insights of Galileo’s physics and Newton’s laws of motion. Hands-on science centres should allow basic laws of physics to be experienced as purely as possible, to relate perception to conceptual understanding. Frictionless toys should be designed and given to babies!
The behaviourists of the 1920s tried to deny consciousness. Most of the phenomena discussed in this book are in our awareness—in our consciousness. Although behaviourism is now a dead doctrine, behaviour itself is of course extremely important, including as an output of visual perception. There would be little point in seeing if we could not act on what we see. What is surprising is how much more there is to perception than appears in behaviour, including appreciation of beauty in nature and the arts.
11.1 Speculative mind-design for vision. Bottom-up signals from the eyes, and other senses, are processed physiologically and interpreted or ‘read’ cognitively by object knowledge (top-down) and by general rules (side-ways). The general rules (such as perspective and Gestalt laws of organization) are syntax; the object knowledge is implicit semantics. Feedback from successes and failures of action may correct and develop knowledge. (Hence the importance of hands-on learning.) It is suggested that real-time sensory signals flag the present—conceivably with qualia of sensation.
Now there is new interest in processes of vision concerned with action. Mortimer Mishkin among others suggests that there are two cortical pathways for vision, one ventral (occipitotemporal, linking striate cortex to prestriate regions, then to inferotemporal cortex); the other dorsal (linking the prestriate regions to the posterior parietal lobe). Interruption of the ventral (‘downwards’) path abolishes visual object discrimination; interruption of the dorsal path from the eyes (‘bottom-up’) disturbs spatial vision, including errors of actions involving positions of objects.
This relates to the distinction between ‘seeing what’ and ‘seeing where’. Brain lesions can affect one and not the other. These brain circuits may also be related to implicit and explicit knowledge (which is hard to define apart from language) and perhaps to what is or not conscious.
There is much of great interest on conscious-visual and behaviour-action pathways in David Milner and Mark Goodales’s (1995), and in Marc Jeannerod’s (1997) recent books, with promise of exciting research to come on relations between mind and brain. No attempt is made in this book to represent anatomical brain systems in the scheme ‘Ins-and-Outs of vision’ (Figure 11.1), as this is quite abstract and so does not reflect brain anatomy or physiology. This awaits further research on relating structures to cognitive functions, with brain imaging with PET scans (page 81), and also deeper theoretical insights into how the brain works. For we must know what functions are involved, to recognize or localize them.
Evidence that there are two visual systems comes from the surprising finding that some distortion illusions are greater as seen than in behaviour. Thus, the two inner circles of the Titchener illusion (Figure 8.10) look different sizes; but reaching and grasping actions are not affected by the visual illusion produced by the different sizes of the outer circles. This separation of seeing from doing has implications for interpreting observations on animals and infants, when language is not available. It makes it even harder to discover what they are seeing from behaviour.
Let’s end with a few thoughts on:
What is consciousness?
Conscious experience—sensations of red, cold, pain, and so on—are our most immediate knowledge and yet are utterly mysterious. In religions and most philosophy, consciousness is supposed to be separate from the brain, which is seen as a receiver of external transmissions. The prevailing view in the brain sciences is very different: that structures and functions of the brain generate mind and consciousness. Francis Crick calls this the ‘astonishing hypothesis’. Historically astonishing and repugnant, this is now what most of us believe. In the mid-18th century the French doctor–philosopher Julien Offray de la Metrie was ostracized and forced to leave his home and work in Paris for just this view, expressed in his book L’homme machine (1748)—yet now this is the received wisdom, at least of the brain sciences.
A striking difficulty is that sensations seem so different from anything physical. It is hard to imagine how physiological activity of brain cells could produce sensations—now often called qualia—such as red or pain. The gap between physiology and sensation seems just too great to bridge by intuitive imagination or by experimental science. But (almost by definition) the most important discoveries of science are counter-intuitive. This why it is unsafe to apply common sense to science, which no doubt can make science frightening.
It is worth pointing out that there are surprising gaps within physics—for example the result of Michael Faraday’s experiment of 1831 which produced electricity by moving a magnet through a coil of wire. Electricity looks incredibly different from moving wires and magnets. Here science gradually developed linking concepts, so this gap is almost filled. Even more puzzling is the timelessness of the equations of physics—yet we know we are living in irreversible time.
In spite of centuries of thought and experiment, gap-filling has not so far been achieved for linking physical brains to mental consciousness. We still hardly know which brain functions are significant. Perhaps even worse, we don’t know how to recognize consciousness in other animals (especially animals very different in looks and behaviour from ourselves), or in a robot. The great difficulty here is that we do not know what—if anything—consciousness does behaviourally.
Here are three hard questions of consciousness:
1. How can physical structure and processes of the brain produce awareness—sensations or qualia?
2. Do qualia have causal effects on brain states or behaviour? (In which case physics and physiology could not be adequate for explaining vision.)
3. If qualia have effects—what do they do?
A dilemma is: if consciousness is causal, we need to incorporate it into accounts of behaviour; if it is not causal, and so no use—why did it evolve through natural selection? It is very hard to see how sensations—qualia—can fit into science if they have no causal consequences, and so cannot be recognized ‘from the outside’. An interesting, controversial move is Roger Penrose’s suggestion to extend quantum physics’ somewhat mind-like characteristics to reduce the gap. If consciousness is at the level of fundamental particles and processes of physics: why isn’t all matter—tables and chairs—conscious? Penrose pin-points certain very small biological structures, microtubules, as specially involved in consciousness; but then we need to ask why (assuming this is so) amoeba are not conscious. In short—what is so special about the brain? It is interesting that our consciousness can be cut off with anaesthetics. If we knew more of how they worked anaesthetics might provide clues, perhaps combined with brain scan experiments.
Whether new techniques for recording brain activity (PET and NMR scans, page 81) will provide basic answers is an exciting possibility. At least they may show where consciousness-generating activity is located in the brain, and they can show which regions are active, not only from sensory stimulation, but according to cognitive processing. There are now criteria for relating physiological brain activity to what it is doing cognitively.
Consciousness is certainly hard to think about. But for the physicist, matter and force and time are perhaps as puzzling, if in different ways. Science is good at describing relations—powerless to handle uniqueness. So perhaps the question ‘What is matter?’ is as hard to answer as ‘What is consciousness?’ because both lack analogies to anything else. It is comforting that consciousness is not the only mind-boggling problem!
May we hazard a guess for what qualia might do? A central notion of this book is that perceptions are but tenuously related to reality; being predictive hypotheses, having major contributions from top-down knowledge and sideways rules derived from the past. Now we may see a problem.
Hypotheses do not especially refer to the present—yet the present moment is crucial for survival. There is no such problem for simple reactions to stimuli—reflexes and tropisms; for stimuli come bottom-up from the present. With development of perceptual hypotheses late in evolution, sensory inputs become relatively less important—as they are greatly enriched by stored knowledge and general rules. Could it be that stimuli retain their original role of signalling the present—but are now flagging the present moment with qualia?
It is suggestive that vision is qualia-rich compared with immediate memory and imagination. Try looking at something, then shut your eyes and imagine it—the difference is surely amazing. The qualia fade in an instant. Memories are far less rich in qualia than perceptions. Could this be because memories lack sensory signals—necessary for signalling the present moment?
An exception is memories of emotions. One may blush with shame, when re-living an embarrassing moment in memory. If the James–Lange theory of emotion is at least in part correct—that emotions are sensations of bodily reactions (page 245)—it could be that remembered emotions trigger qualia from the sensed bodily changes. Qualia of vivid dreams, and schizophrenic hallucinations, might be where this supposed qualia-flagging system for identifying the present breaks down. Conceivably this has clinical implications. This qualia idea is speculative; but as the tortoise said, ‘I can’t take a step forward without sticking my neck out.’
A heavily top-down robot would need to use its real-time inputs to signal the present moment; but would it need qualia? Should we ever come to confer our processes of perception to machines, perhaps problems of consciousness will be solved. Or, perhaps we need to answer qualia questions to be able to build conscious machines. Meanwhile, as for any science, we may continue to study and try to explain phenomena without ‘ultimate’ understanding. This is how science works, as any explanation has limits.
The physical sciences take immense trouble to avoid errors. Here we seek out and study errors for understanding how we see and to suggest something of how the brain works. The weird and wonderful errors of illusions are not trivial. They are truly phenomenal phenomena, central to art and a major reason for the experimental methods of science.
