Framing the indefinable

We encounter science as a natural kind, not as an abstract category. In other words, like a chair, or a tiger, or a city, we recognize it when we come across it, without having to refer to an explicit formula. Indeed, such a formula is not feasible. It would not only have to be elaborate enough to indicate that science has many different aspects – institutional, mental, material, and so on. It would also have to be broad enough to extend over many different instances of scientific activity, from classifying beetles to theorizing about black holes, from recording folk tales to mapping the human genome, from ancient Chinese medicine to modern Japanese pharmacology, from explaining earthquakes to failing to explain inflation.

A catalogue of all these aspects and instances would obviously be quite unmanageable. It would merely demonstrate that science is too diverse, too protean, to be captured in full by a definition. Moreover, any such definition would pre-empt the outcome of our enquiry. By telling us in advance what science is, it would effectively determine what we would later surely find. We may be very well informed about science, and have a very good idea of various features that are typical of it, but we must be careful not to insist that any of these features are invariable or definitive.

Understanding and explanation

Science produces knowledge. This is something more than codified information. As we have seen, the notion of ‘knowledge without a knower’ [9.3] cannot be taken literally. The myriads of facts and theories in the scientific archives have been shaped by the requirements of interpersonal communication. They have to be meaningful: they have to be capable of being understood.

This meaning may only apply in a very esoteric context. The necessary understanding may be limited to a tiny, highly specialized research community. The production process may have involved an opaque conglomeration of automatic instrumentation, computation and symbolic manipulation. Nevertheless, the norms and practices of academic science require that the nature of this process and its final products should have been communicated to and consciously accepted by human minds. As I have repeatedly stressed, the epistemology of science is inseparable from our natural faculty of cognition.

But scientific communications must not only be comprehensible: they also, typically, enable comprehension. For reasons that we have discussed at length, they relate directly or indirectly to shared aspects of the life experiences of those who utter and receive them. In a word, they are messages that we send to each other about the ‘world’ that we seem to have in common. They thus help us to understand that world.

‘Understanding’ is a complex process, of which we know less about the parts than about the whole.

Problems

Academic science is energized by the norm of originality [3.6]. This norm requires scientists to produce new knowledge – that is, communally acceptable information that was not previously known. To do this, they engage in research. But they can be credited with CUDOS [3.8] only for what they discover through research that they have themselves decided to undertake. Academic science – and, in the end, cognitive change – is thus steered by innumerable independent decisions of this kind.

The significance of this process is often under-estimated. Philosophers constantly insist that scientific knowledge is provisional, but they seldom remind us that, even though it is continually expanding, it is very patchy in its coverage. This is not just a regrettable weakness, which can be forgiven because it will in due course be made good. It is a fundamental epistemic characteristic of academic science, closely connected with its social structure and cultural practices.

To put it simply: at any given moment, what we know and how well we know it depends on what our predecessors decided to study in the past. What we shall know in the foreseeable future will depend on what research we undertake now. For example, we would not nowadays be studying the medical uses of genetic information if Francis Crick and James Watson – not to mention many others – had not individually decided, nearly 50 years ago, to investigate the structure of DNA.

The academic mode

Science is a mode of knowledge production. Its social norms are inseparable from its epistemic norms – what philosophers call its regulative principles. Scientists' ideas about what should count as ‘the truth’ cannot be disentangled from the ways they work together in pursuing it. The philosophy of academic science is part and parcel of its culture.

The regulative principles of academic science are thus important components of its ethos. They are actually so familiar to most scientists, and are stated so often in different forms, that it is not easy to produce a standard list. The simplest way of describing them is to say that they involve such concepts as theory, conjecture, experiment, observation, discovery, objectivity, inference, etc., which we shall be analysing in detail in later chapters.

The significant point here is that although these are usually taken to be independent philosophical concepts they can be directly linked with sociological aspects of the academic ethos. For example, the norm of ‘communalism’ is closely connected with the principle of empiricism – that is, reliance on the results of replicable observation and experiment. Again, social ‘universalism’ is related to explanatory unification; ‘disinterestedness’ is normally associated with belief in an objective reality; insistence on ‘originality’ motivates conjectures and discoveries; ‘organized scepticism’ requires that these be fully tested and justified before being accepted as established knowledge. And so on.

This correlation with the Mertonian norms accords with our naturalistic approach.

The seeds of this book were sown forty years ago. I was always infatuated with science and beguiled by philosophy. They seemed made for each other – and for me. But the better I came to know science, the more I realized that the philosophers were not telling it like it is. Then, sometime around 1959, I was asked to review Michael Polanyi's Personal Knowledge and Karl Popper's The Logic of Scientific Discovery. Each of these great books says important things about science; but in both I noticed a whole pack of dogs that didn't bark. What about the web of lectures, examinations, seminars, conferences, papers, citations, referee reports, books, personal references, job interviews, appointments, prizes, etc. in which my scientific life was entangled? Surely these must have some influence on the work I was doing. So in radio talks and articles I began to say strange things, such as ‘Science is social’ and ‘Research is a profession’.

Those were rash words for a young and aspiring physicist without official credentials in philosophy or sociology. Nevertheless, the heterodoxy was overlooked and my academic career prospered. The books in which I developed this theme – Public Knowledge, Reliable Knowledge and An Introduction to Science Studies – were also very well received, and are still read and cited. Indeed, many of the notions that germinated in these books have since been planted out more formally by other scholars. And just as I foresaw, sociology has superseded philosophy at the theoretical core of ‘science studies’.

The agonistic element

The final norm of academic science [3.7] is organized skepticism. This sounds like a philosophical doctrine, but is not a call for total doubt. The metaphysical notion that we cannot really know anything is not incompatible with being a scientist, but is so general and abstract that it has no more impact on scientific practice than it does on other aspects of life [8.10, 10.5]. Again, scepticism has psychological overtones, favouring a ‘questioning’ attitude, akin to ‘curiosity’ [2.7]. This attitude is as necessary to scientific progress as personal ‘creativity’, although it must not be confounded with a conservative stance that automatically rejects every new idea.

But its real force is sociological. ‘Scepticism’ is a code word for those features of the scientific culture that curb ‘originality’. Personal trust is an essential feature of scientific life [5.7]. But scientific communities do not accept research claims on the mere say-so of their authors. The active, systematic exercise of this norm by individual researchers is what, above all, makes science a communal enterprise. ‘Peer review’ is the key institution of the scientific culture.

We have already noted a number of the ways in which this norm indirectly shapes scientific knowledge. To be considered ‘scientific’, a ‘fact’ or theory has to satisfy a number of general epistemic criteria, such as reproducibility, logical consistency, independence of the observer, etc. These are essential conditions for communal acceptability.

Defending a legend

Science is under attack. People are losing confidence in its powers. Pseudo-scientific beliefs thrive. Anti-science speakers win public debates. Industrial firms misuse technology. Legislators curb experiments. Governments slash research funding. Even fellow scholars are becoming sceptical of its claims.

And yet, opinion surveys regularly report large majorities in its favour. Science education expands at all levels. Writers and broadcasters enrich public understanding. Exciting discoveries and useful inventions flow out of the research laboratories. Vast research instruments are built at public expense. Science has never been so popular or influential.

This is not a contradiction. Science has always been under attack. It is still a newcomer to large areas of our culture. As it extends and becomes more deeply embedded, it touches upon issues where its competence is more doubtful, and opens itself more to well-based criticism. The claims of science are often highly questionable. Strenuous debate on particular points is not a symptom of disease: it signifies mental health and moral vigour.

Blanket hostility to ‘science’ is another matter. Taken literally, that would make no more sense than hostility to ‘law’, or ‘art’, or even to ‘life’ itself. What such an attitude really indicates is that certain general features of science are thought to be objectionable in principle, or unacceptable in practice. These features are deemed to be so essential to science as such that it is rejected as a whole – typically in favour of some other supposedly holistic system.

Striving towards objectivity

When Robert Merton pointed out that scientists are constrained to be ‘disinterested’, he was referring mainly to their professional behaviour. As we have seen [3.5], there are many conventions that severely limit the operation and public display of personal motives in the regular practice of academic science. The direct effect is thus to sever the connections between scientific knowledge and its personal origins.

According to this norm, it might seem that scientific knowledge should always be presented as cognitively objective – i.e. as if referring to entities that exist quite independently of what we know individually about them. But the implication that science rests on and requires absolute realism calls for further discussion, which we postpone to a later chapter [10.7].

As a social norm, however, disinterestedness functions primarily to protect the production of scientific knowledge from personal bias and other ‘subjective’ influences. Strictly speaking, this is impossible [5.3]. There is no denying that scientific facts and theories are produced by human beings, whose minds cannot be completely cleansed of individual interests. Academic science therefore strives to attain consensual objectivity by merging these interests in a collective process.

The norm of disinterestedness thus combines naturally with the norms of communalism and universalism to strip scientific knowledge of its subjective elements and turn it into a genuinely communal product. Indeed, the impersonal style of formal scientific discourse is designed to make research claims appear immediately acceptable.

Generalization and abstraction

The academic ethos lays down that scientific knowledge has to be the common property of a universal community [3.4]. Social realities obviously limit the scope of this norm. Nevertheless, it firmly shapes the type of knowledge that is admitted into the scientific archive. In principle, science deals only with what could be communicated to and accepted by anybody, regardless of their other personal beliefs or special circumstances.

The result is that science is primarily concerned with generalities. Particular facts are not specifically excluded from the scientific archive. Many serendipitous discoveries – especially in the biomedical sciences – have been triggered by published reports of apparently singular events. But such particularities are very, very seldom of ‘universal’ interest. It is quite impracticable to ‘share’ with other scientists the immense quantities of factual information that accumulate in the course of research. If this information is to become communal property, it must be encoded and ‘compressed’ into a much more compact form. In other words, the detailed facts must be interpreted and presented as specific elements of more general patterns – typically as entities governed by theories.

Trying to give a basic, comprehensive account of the concept of a ‘theory’ is an invigorating but fruitless walkabout in metaphysics. In all that follows, we shall treat ‘theories’, like ‘facts’, as primitive epistemic entities – ‘natural kinds’ as some philosophers call them – with which we are better acquainted from experience than from philosophical analysis. Indeed, from this naturalistic point of view, facts and theories are closely interwoven. It is impossible to talk about knowing something as a scientific fact without reference to a theory [5.5, 5.6].

The republic of learning

Academic science is the stereotype of science in its purest form. When people talk about scientific research (as distinct from technology) they primarily have in mind the sort of scientific work that is done in universities. They think of it as the characteristic activity of members of a particular social group in a particular social frame.

Scientists themselves insist that they belong to a community, indicating that they recognize each other as people who share many values, traditions and goals. But this community is essentially notional. The word is used to mean ‘all those people who subscribe to certain general principles of rationality and objectivity, and have such high standards of expertise and mutual trust that they can be relied upon to work together for the benefit of humanity in the attainment of truth’. On the one hand, it proclaims the unity of this group within society at large. On the other hand, it asserts that its members are individuals who are linked together voluntarily by their common attitude to learning and research.

The concept of a scientific community is part of the traditional philosophical Legend. At the same time, however, it encases science in a sociological ‘black box’, whose internal structure is deemed to be irrelevant to the pursuit of knowledge. Indeed, the power of the Legend lingers on, even amongst the champions of a ‘sociology of scientific knowledge’.

Oral and literature cultures

In approaching this topic, I decided to start by discussing memory in oral cultures, which is what I call those without writing. Unlike many other scholars, I use the phrase ‘oral tradition’ to refer to what is transmitted orally in literate cultures. The two forms of oral transmission in societies with and without writing are often conflated, and that has been the case in the well-known work of Parry and Lord on the ‘orality’ of Homer. Most epics are products of literate cultures even if they are performed orally.

Oral performance in literate societies is undoubtedly influenced to different degrees by the presence of writing and should not be identified with the products of purely oral cultures. The point is not merely academic for it affects our understanding of much early literature and literary techniques, which are seen by many as marked by the so-called oral style. To push the point to a speculative level, speculative since I do not know a sufficient number of unwritten languages (and here translations are of no help whatsoever), many of the techniques we think of as oral seem to be rare in cultures without writing. Examples include assonance (as in Beowulf or the work of Gerard Manley Hopkins), mnemonic structure (as in the Sanskritic Rig-Veda), formulaic composition and even the very pervasive use of rhyme.

In his Vie d'Henri Brulard, the French novelist Henri Stendhal articulated the difficulties faced by an autobiographer endeavouring to recapture his life of thirty years earlier: ‘I make many discoveries … They are like great fragments of fresco on a wall, which, long forgotten, reappear suddenly, and by the side of these well-preserved fragments there are … great gaps where there's nothing to be seen but the bricks of the wall. The plaster on which the fresco had been painted has fallen and the fresco has gone forever.’ The richness of human life depends on our ability to remember the past, yet – like Stendhal – we are painfully aware of our memory's selectivity and vulnerability. Whether we are trying to recall particular details of our own lives, or to construct historical narratives describing broader cultural changes, we must all confront the gaps and distortions inherent in recapturing the past. This perplexing faculty, so central to our existence, exerts a universal fascination: as individuals, we wish to learn more about how our own memory functions; as members of society, we are concerned to appreciate the multiple ways in which our history is preserved. What we remember is intimately linked to how we remember, but innumerable approaches have been devised to explore that complex web of connections. The eight chapters in this volume, all by leading experts in their fields, transcend this diversity to address together the relationships between individual experience and collective memory.

Although she came to see me as an analytical patient five times a week, Miss A. found it very difficult to remember anything from one session to the next. Shortly after we had met and started working together, she had told me that she needed me to remember why she had gone into a shoe shop. It was not like going into a grocer's and forgetting the sugar – anyone could do that; she did not even know why she was in the shoe shop. She added, somewhat embarrassed by what she had said, that she did not need me to know it too well, that would be absolutely awful.

Mr B., on the other hand, had an excellent memory and prided himself on recalling what I had said more accurately than I did. But one day when I mentioned something it turned out he had forgotten, he snapped at me, ‘How can you expect me to remember that when I can't even remember my own mother from one day to the next?’ Mr B.'s widowed mother needed his daily attention. Mr B. had a particular memory of childhood that was probably what we call a ‘screen’ memory – a mixture of experience and highly relevant fantasy, like a dream image. In this mnemic image he was standing shaking out a tarpaulin sheet together with his father, a DIY enthusiast: he could see the garden, himself, his hands on the tarpaulin, the tarpaulin held at both ends shaking, and then the image stopped.

The brain's operation depends on networks of nerve cells, called neurons, connected with each other by synapses. Scientists can now mimic some of the brain's behaviours with computer-based models of neural networks. One major domain of behaviour to which this approach has been applied is the question of how the brain acquires and maintains new information; that is, what we would call learning and memory. Neural networks employ various learning algorithms, which are recipes for how to change the network to store new information, and this chapter surveys learning algorithms that have been explored over the last decade. A few representative examples are presented here to illustrate the basic types of learning algorithms; the interested reader is encouraged to consult recent books listed in the section on Further reading, which present these algorithms in greater detail and provide a more complete survey.

It is important to keep in mind that the models of neural networks that are simulated in computers are far simpler than the highly complex and often messy neural systems that nature has devised. Although much simpler, neural network models capture some of the most basic features of the brain and may share some general properties that will allow us to understand better the operation of the more complex system. Neural network models are built from simple processing units that work together in parallel, each concerned with only a small piece of information.

Although all of us experience memory failures at some time or another, these slips of memory do not cause severe disruptions to our daily lives. Most of us can still function adequately at work, engage in conversation, and remember the gist of the programme we saw on television last night, while accepting as normal the forgetting of certain details. After all, nobody remembers everything. For some people, however, their memory failure is of such a proportion that the effects can be devastating.

Imagine waking up and not being able to remember what you did yesterday. Imagine living in a time vacuum where there is no past to anchor the present and no future to anticipate. Such is the fate of many people suffering from organic amnesia. Although amnesia means literally ‘an absence of memory’, in practice people with organic amnesia do not have a total loss of memory. They remember who they are, they remember how to talk, and how to read, and they usually remember how to do things they learned before the onset of their memory loss, such as how to swim, ride a bike or play the piano. Unfortunately, they have great difficulty in learning new skills or information, experience problems when trying to remember ongoing events, and usually have a memory gap for the few days, weeks, months or even years before becoming ill.

In contrast, people with functional amnesia following, say, an emotional trauma, sometimes seem to lose memory for personal identity.