3 Contributions of neuroscience to philosophy
Once the neurosciences began to investigate the neuronal underpinnings of higher cognitive functions, especially those realized in human brains, an increasing number of questions, traditionally investigated by the humanities, were addressed through empirical studies within the rapidly developing field of cognitive neuroscience. One obvious domain for this investigation was epistemology. Cognitive neuroscience explores from a third-person perspective the mechanisms that mediate our perception and the acquisition of knowledge. Longstanding discussions about the objectivity of cognition, the question of how constructive our perceptual processes really are, and how reliable or idiosyncratic they might be, need to be reconsidered on the basis of neurobiological data. Another question, to which the neurosciences will have to find an answer, is related to the mind-body problem: how can mental phenomena, namely immaterial entities such as the qualia of perception and social realities such as belief and value systems, emerge from the material interactions between nerve cells in human brains? These immaterial phenomena came into this world once the cognitive abilities of Homo sapiens initiated the evolution of cultures. They affect our lives as much as the material constraints of the world in which we evolve, but they have a different ontological status to the neuronal processes that brought them into this world. Yet another question that solicits discussion between neuroscientists and philosophers of mind is the nature of consciousness. The question of the constitution of the intentional Self is closely related to this issue, as is the conundrum of the existence of Free Will. If the material processes in individual brains and the social realities resulting from the interactions of humans are the basis and cause of mental phenomena, and if brain processes follow the known laws of nature, then there ought to be unifying description systems that bridge the gap between phenomena assessed from third- and first-person perspectives. If such approaches turn out to be feasible philosophical positions, the postulation of an ontological dualism will have to be modified. This will have far-reaching consequences for our self-understanding and the delineation of the border between “physics” and “metaphysics”.
3.1 An epistemic caveat
It is obvious that our perceptions and imaginations, as well as our ability to reason, are constrained by the cognitive abilities of our brains—and brains, like all other organs, are the product of an evolutionary process. Hence our brains have become adapted to the conditions of the mesoscopic world in which life has evolved. This is the world within the scale of millimeters to meters, it is the world where the laws of classical physics prevail; it is not the world of quantum physics and it is not the world of astrophysics. As a consequence, our cognitive functions have become adjusted to assure survival in this mesoscopic world. Problem-solving in this dangerous and poorly-predictable world requires the application of pragmatic heuristics and hence cognitive abilities that are in all likelihood not optimized to comprehend the essence behind the perceivable phenomena or the “absolute truth” in the Kantian sense. Evolution did not prepare us to directly perceive and understand processes at subatomic or cosmic scales, because they were and are completely irrelevant for our daily struggle for survival. Even more worrying is the possibility that the way in which we reason may also be limited by adaptation to those processes in the narrow range of the world that are relevant for survival and that we can access with our highly selective, specialized senses. In conclusion, it is very likely that our cognition is constrained. And this may apply not only to primary perception, but also to our way of deriving inferences from observables. If this were true it would pose unsurmountable barriers to our attempts to understand, just as it would challenge the consistency of mathematical theories and logical deductions. However, for these very reasons we have no way of knowing whether this is the case.
3.2 The contribution of neuroscience to epistemology
Growing insights into the neuronal mechanisms underlying perception provide compelling support for constructivist positions and emphasize the epistemic caveats formulated above. In the light of neurobiological evidence, perceiving is essentially a constructive process. The sensory categories, for example those according to which we assign qualities to our experiences, are nothing but the idiosyncratic consequence of the layout of our sensory organs. These sample in a highly selective way a narrow range of physico-chemical signals, and this leads to the arbitrary classification of electromagnetic radiation with wavelengths between 400 to 700 nanometers as light, because the photoreceptors in the eye are sensitive to this wavelength range. Radiations with slightly longer wavelengths stimulate our temperature receptors and we categorise the respective sensations as temperature. A similar arbitrariness of category boundaries is observable in other sensory domains. The definition of perceptual objects, for example, is guided by a set of Gestaltrules that our brains apply in order to segment the spatio-temporal continuum of sensory signals into distinct objects—and this holds true for all sensory modalities. Objects are identified as such if they are delineated by spatial or temporal borders and exhibit some intrinsic coherence. This definition is appropriate in the mesoscopic world, but it does not apply to objects at atomic or subatomic scales. If we had no a priori definition of the properties of objects, we would not be able to distinguish objects, we would, for example, be unable to extract object-specific features from the continuous two-dimensional brightness distribution that cluttered scenes generate on the retina.
It is now well established by experimental evidence that the sparse sensory signals provided by our highly selective senses are interpreted by the brain on the basis of a vast amount of a priori knowledge that is stored in its own functional architecture. Our self-active brains permanently formulate knowledge and context-dependent expectancies, interpret sensory signals as a function of these inferences, and present the result of this constructive process to the workspace of consciousness. Paradoxically, we perceive the world around us as coherent even though our senses extract only a minute fraction of the available signals. Much of what we experience as actually perceived is read out from memory and is the result of reconstruction and completion. This raises the question of the origins of this knowledge.
3.3 The sources of a priori knowledge
It is commonly accepted that all the knowledge a brain can possibly have, and the rules according to which this knowledge is applied for the interpretation of sensory signals and the execution of movements, reside in the functional architecture of the brain. This contradicts the analogy frequently drawn between computers and brains. Computers have processors and separate memories for programmes and for data. In the brain, however, there exist only neurons and connections. Both the stored knowledge and the programs for processing this knowledge reside in the layout of these connections, their polarity—that is, whether they are excitatory or inhibitory—and their graded efficacy. The question of the origin of stored information is thus reduced to the question of which processes determine the functional architecture of the brain.
The most important determinant of the functional architecture of brain—and hence the most important source of knowledge—is, of course, evolution. What makes our brain architectures comparable is evolutionary-acquired information that resides in the genes and determines the layout of the brain’s connectome. It is knowledge about the world that is expressed in the functional architecture of brains every time an organism develops. In this sense evolution can be considered a cognitive process. This evolutionary-acquired knowledge pertains essentially to the conditions of the precultural world; and it is implicit—we are not aware of having it because we were not around when it was acquired. Still, we use it to interpret the signals provided by our sense organs and to structure adapted responses.
This inborn knowledge is subsequently complemented by extensive epigenetic shaping of the neuronal architectures, which adapt the developing brain to the actual conditions in which the individual develops. The human brain develops the majority of its connections only after birth, and this process continues approximately until the age of twenty or twenty-five years. During this developmental period numerous new connections are formed, while many existing connections are removed; and this making and breaking is guided by the neuronal activity itself. Since, after birth, neuronal activity is modulated by interactions with the environment, the development of brain architectures is thus determined by a host of epigenetic factors derived from the natural and social environment. Through this process, the brain acquires knowledge about the specific conditions in which the newborn organism actually evolves, and thereby complements its genetically-inherited knowledge.
A considerable part of this developmentally-acquired knowledge also remains implicit because of the phenomenon of childhood amnesia. Children before the age of about four years have only a limited capacity to remember in which context they have experienced and learnt particular contents. The reason for this is that the brain centres required for these storage functions—we call them episodic or biographical or declarative memories—have not yet matured. Thus, while young children learn very efficiently and store contents in a very robust way through structural modifications of their brain architecture, they often have no recollection of the source of this knowledge. Because of this apparent lack of causation, the knowledge acquired in this way is implicit, similarly to evolutionary-acquired knowledge, and therefore often assumes the status of convictions that cannot be questioned.
Like innate knowledge, this acquired knowledge is used to shape cognitive processes and to structure our perceptions. Yet we are not aware that what we perceive is actually the result of such knowledge-based interpretations.
Finally, there is knowledge acquisition by learning, which accompanies us throughout our lives. This is based on graded changes of the coupling strength of the existing connections between neurons. In the adult brain few new connections are formed and under normal conditions no breaking of connections occurs. The knowledge acquired by these learning processes also biases perception, but this is explicit, and its origins are known. One is usually aware of when and how it has been acquired and can therefore question its validity and by the same token the validity of what is perceived.
3.4 Examples illustrating the influence of priors on perception
The two examples depicted in figure 1 and figure 2 illustrate impressively how a priori knowledge structures our primary perceptions. The object in figure 1 is a mould used to produce candies. On the left side one sees the inside of the mould, with its concavities, and on the left we see the rear side with its corresponding convex protrusions. In reality, the pictures are identical, but one is rotated by 180°. The reason for our very different perceptions of the images is that the brain makes the a priori assumption that light comes from above—a well adapted assumption in a precultural world with only natural light sources. In this case contours that have the shadow above are interpreted as concave, and those with the shadow below as convex. Thus, an assumption of which we are not aware determines what we perceive. Another striking example is shown in figure 2. It is hard to believe, but surfaces A and B have exactly the same luminance. They appear different because the brain sees the shadow that is caused by the cylinder on the right. Even though the amount of light reflected from surfaces A and B and impinging on the retina is exactly the same, the brain interprets the brightness of the two surfaces as different because it infers the following. Given that there is a shadow, surface B must be brighter than surface A—which has no shadow on it—in order to reflect the same amount of light. Thus, the brain “computes” the inferred brightness of the surfaces, but we are not aware of these computations. We just perceive the result and take it to be real, i.e., we see B as being much brighter than A. These two examples indicate that the brain generates inferences of which we are not aware, that it is permanently reconstructing the world according to a priori knowledge, and that we, as perceiving subjects, have to take for granted what the system finally offers us as conscious experience. As expected, this is not only the case with specially designed psycho-physical experiments, but is an essential feature of all our perceptual processes.
Figure 1: The brain assumes that light comes from above. The circular contours on the left board appear as concavities because the shadows are located at the right upper corner. The right board is actually the same as on the left, just rotated by 180 degrees. Now the shadows are on the lower left border and the contours appear convex.
Figure 2: The checker-board illusion by Adelson, illustrating that even brightness perception depends on assumptions derived from context. (For further descriptions see text.)
The mechanism leading to this “false” perception, to this “illusion”, has of course an important function. Our brain uses this principle to generate perceptual constancy, e.g., to keep colours and contrasts constant despite different illumination conditions. The spectral composition and the intensity of the sunlight change dramatically throughout the day, and therefore the spectral mix of light reflected from a particular, edible berry differs in the morning from that at noon. An animal that relies on colour to distinguish one edible berry from another, slightly more violet and poisonous berry, cannot rely on an analysis of the “true” or actual spectral composition of reflected light. It first has to assess the spectral composition of the light source—the sunlight—and then must reconstruct the perceived colour. Our brains accomplish this by assessing the actual lighting conditions, by comparing the colours of the sky, of stones, of leaves and barks etc. and then, by using this contextual information, compute the “real” colour of the berries to identify that which is edible. Our brains are capable of assuring colour constancy despite changing illumination conditions, but we are completely unaware of the complexity of the computations assuring constancy and thereby survival in a changing world. In essence, all these operations are based on the evaluation of relations. We rarely perceive absolute values such as those measured by physical devices, be it intensities of stimuli, wave-lengths of sound or light waves, or chemical concentrations. We mostly perceive these variables in relation to others, as differences, increments and contrasts, such that these comparisons are made both across space and time. This is a very economical and efficient strategy because it emphasizes differences, permits coverage of wide ranges of intensities and, as mentioned above, allows for constancy. Given the advantages of these well-adapted heuristics it is at least questionable whether one should confront the resulting perceptions as illusions.
3.5 Conclusion of the excursion into epistemology
Evolution- and experience-dependent development determine and shape the architecture of the brain. Through these processes, knowledge about the world and strategies to use this knowledge for survival and reproduction are implemented in brain architectures. These in turn determine what and how an organism perceives and how it behaves. Because of the selection criteria that guides evolution, the brain adapted to the narrow segment of the world in which life has evolved, and its functions have been optimized to extract and process those signals that best serve survival and reproduction. Thus the cognitive functions of the brain have probably not been optimized for understanding the deeper structure of the world that assures coherence across scales and cannot be perceived directly. Similarly, the rules according to which we evaluate contingencies and establish associations among events are implemented by specific molecular mechanisms that translate temporal correlations of neuronal activity into lasting changes in the efficacy of neuronal connections. These rules have been preserved virtually unchanged since the evolution of primitive nervous systems, and are at the basis of assignments of causality and the formation of associations. Again, these rules are highly efficient for the generation of models of the mesoscopic world and the formulation of predictions, but they do not apply to processes in the quantum world or to the relativistic dynamics of the universe. Given these specific adaptations of our cognitive functions, one might consider that similar restrictions may also hold for the way we reason. If so, this would present a serious challenge for the generalisation of models and theories based on extrapolation.