welcome covers

Your complimentary articles

You’ve read one of your four complimentary articles for this month.

You can read four articles free per month. To have complete access to the thousands of philosophy articles on this site, please


Knowledge & Reasons

Joe Cruz gives an evolutionary account of them.

Epistemologists are philosophers who seek to understand what knowledge is, whether we have any of it, and whether, perhaps, we can improve our ability to gain more of it. ‘Knowledge’ is understood here in a general sense. Particular bits of knowledge – for instance, knowing that 7+5=12, or how to knit mittens, or that George Washington was the first President of the United States – are treated as data. Epistemologists ask, what makes possession of each of those bits of data instances of knowledge, as opposed to merely opinion?

There is a difference between having knowledge and merely having an opinion even if the opinion is correct or useful. After all, one could have a true opinion accidentally, by, say, correctly guessing. If I am lost at a fork in the road and flip a coin to decide which way to go, it may turn out that the opinion I form leads me to where I wanted to go. Still, I did not know that going left was correct, even though it was. So being true does not by itself make a belief properly knowledge – although most philosophers think that being true is a necessary condition for knowledge. Moreover, having opinions, even useful, true ones, does not tell us what procedure to employ to get new useful, true beliefs. Opinions are in a sense inert. By their nature, mere opinions are arrived at in a way that is uncertain and obscure. It is only by understanding knowledge, epistemologists think, that we can obtain an intellectual procedure with which we can move forward . So epistemologists set out to determine what is needed to make the difference between a true opinion and knowledge.

One natural thing to think, is that in addition to being true or useful, knowledge requires having good grounds, reasons or evidence for what is known. This is what was missing in the coin flip at the fork in the road: I had no reason to think left was correct, even if it turned out to be. Knowledge and rationality therefore seem intimately connected, in that having knowledge requires good reasons, and being rational is the process of employing good reasons. Epistemology has thus expanded to include the study of the rationality or reasonableness of beliefs, and it has done so in a way that suggests a close relationship between knowledge and rationality.

Reasoning About Reasons

Unfortunately, the introduction of the need for good reasons trades one mystery for another. What makes some reasons good and other reasons bad? To answer this question epistemologists will need to give an explanation of what features any reason has to have in order to be good.

Certain kinds of answers to this question are tempting but probably cannot give a fully general theory of good reasons. For instance, mathematicians have a decent grip on what counts as a good reason in a mathematical proof: a reason is good in mathematics if it is sanctioned by the rules of the formal system in question. So it is reasonable to believe that 7+5=12 – in fact, we may even know it – because the answer is determined by the rules of arithmetic within a fixed notation system. But this does not provide a general answer to the question of good reasons, because most areas of thought do not have tightly-defined rules and fixed meanings of the mathematical sort. Still, the mathematical case might be illuminating. What is promising about it is that there is an acceptable starting point for good mathematical reasoning – the axioms – and an acceptable procedure for drawing conclusions – the rules. If there is something like that in the non-mathematical domains, then we might be in a better position to understand good reasons both in ordinary life and in science.

Historically, and intuitively, perception has seemed a good starting point for forming rational beliefs. If someone sees or hears or otherwise perceives something, we think that generally, that is an acceptable basis for her to form beliefs about that thing, and for those beliefs consequently to be rational. To be sure, there are instances where we would not think this. If someone has reason to suspect that her perceptual experience is distorted or otherwise unreliable, then we would not think that her beliefs formed on the basis of that perception are reasonable. But generally as a starting point, it seems that perception is an acceptable basis for forming rational beliefs. Of course, to be reasonable, those beliefs have to be about the thing perceived. If I see a chair in front of me, that is not by itself a good reason to believe that George Washington was the first President of the United States. Even bringing up George Washington is absurd, because the US presidency has nothing to do with chairs – barring some convoluted and circuitous connection that would need to be spelled out.

This gives us some sense, then, of the rules that govern the formation of rational beliefs on the basis of perception: (pending some good reason to think otherwise) if someone forms a belief on the basis of the perceptions she has, and the belief is about the perceived properties of the thing, beliefs so formed are rational. The model of rationality suggested by mathematics, that is, the idea of having a fixed starting point and having rules that determine what beliefs are acceptable to form on the basis of that starting point, has now been applied by analogy to a non-mathematical domain. Many epistemologists think that, if a similar account can be given for a range of beliefs that are not about perception – for example, ones about memory or testimony or probability – then we will have made headway on the main project of epistemology, to account for knowledge generally.

Problems With Reasons

There is an important wrinkle in this theory of good reasons. It is ambiguous to whom the reasons have to be good in order for beliefs formed on their basis to be rational. That is to say, my belief may be rational to some third party if they were to examine it, but if I myself do not appreciate the reasons for it, it is plausible to say that I am not rational in believing it. Imagine that I complete a maths quiz by fudging some steps. I have no idea whether the steps are legitimate, but suppose, unbeknownst to me, I accidentally get those steps right. Even though my conclusions are correct, they are not rational, because I did not appreciate the good reasons leading to the conclusion. Rationality thus appears to have a strong first-person dimension, in that the believer must possess the evidence and the rules for drawing rational conclusions herself, and must in some way appreciate that fact as she applies the rules.

There will be many ordinary cases where a person does not have a particularly good grasp on the reasons for a belief she has. She may have forgotten her reasons, but still feel fairly confident in the belief. Or she might recognize her reasons, but not be able to articulate what makes those reasons good. If cross-examined by a skeptic, she may be unable to defend herself.

If confronted by a very persistant skeptic, to uphold her rationality a person may need to say not only what her good reasons are for believing that the reasons she has are good, but what her good reasons are for believing that those underlying reasons are good, and so on. No one can give good reasons for good reasons for good reasons indefinitely, so for it to be possible for anyone to be rational, there must exist legitimate stopping points in reason-giving. One need not even be able to get back to those stopping points in every case. At the very least, however, we expect a person to be able, if allowed time and patient reflection, to recover in some way what her reasons for her beliefs are, and very broadly, why they are good ones.

Here is where we are so far. In order to be knowledge, a belief has to be true and based on good reasons for it. Good reasons are different from bad reasons in that good reasons can be traced to some source such as perception that keeps us connected with the thing that we form the belief about.

These modest bits of progress can certainly be challenged and rejected, and, even if they are right, they invite a dizzying list of further questions. Contemporary epistemology is largely an attempt to give reasoned, systematic answers to such questions. But we should notice that as interesting, challenging, and worthy as it is, this agenda of refining this framework of epistemology is something akin to writing the final chapter in a much broader epistemological narrative. It is equally important to see how philosophers conceive of the even bigger picture: How is it that the capacity to know things is present in human beings? What makes knowledge possible at all? This is the question I want to address in the rest of this article.

The Automatic Behavior of Living Things

Nonliving things remain what they are because they passively resist forces that would compromise their integrity and persistence. (By ‘integrity’, I mean being a thing staying together spatially, and by ‘persistence’, I mean a thing with a continuous intelligible nature enduring over time.) By contrast, the persistence and integrity of living things is achieved actively – by their doing something to adapt to their environments; to respond to challenges or to change their environment to increase their favorability.

The distinction between passive and active here is metaphysically problematic. What could it mean to say that some things ‘don’t do’ and some things ‘do’? In a sense, everything does as it does. We see the strain involved in making the distinction between non-living and living things, and therefore between passive and active, in, for example, viruses. However, scientists are not usually vexed about the definition of the boundary. Instead, scientists mark the distinction by offering different ways of thinking about nonliving and living things. The central domain of inorganic chemistry is nonliving things, while the central domain of biology is living things. The border cases provide fascinating and lively areas of scientific inquiry. Organic chemistry can be viewed as a science of the border cases that forms a bridge between understanding the nonliving and the living. It’s a good thing, since it shows that explanations at various levels of analysis can be reconciled, by saying that biological things are made of chemical things that are in turn made of physics things.

Let us use the explanatory (as opposed to the metaphysical) sense of the ‘passive’ versus ‘active’ distinction to sharpen our broader epistemological picture. Consider the contrast between rocks and ants. Rocks have a physical nature (eg, their solidity) that passively preserves what they are against many kinds of influences in the world. There are some changes that rocks cannot retain their integrity and persistence against, such as an earthquake or the flow of a river, and so rocks might be transformed into something else passive, like dust. For a span, however, rocks are what they are because of a fixity that passively resists change. The vast majority of things in this universe seem to stay what they are in the same sort of way.

Ants are different. They actively cause changes in their environment in order to maintain a certain equilibrium with that environment. They engage in adaptive, effortful activity to preserve their nest, for example. Ants have a fixed nature too, but this nature has a different character to that of rocks. Rocks are persistent wholly in virtue of a relatively static structure, whereas ants wouldn’t remain the same unless they could change. An ant is an organism evolved to intervene actively ‘on its own behalf’, and it does so by dint of the flexible arrangement of its parts. Ants’ parts self-organize, shifting their relationships to one another. It looks like this activeness disappears under analysis, since it seems like it’s achieved through the complex interaction of parts that are individually passive. So the physical makeup of ants, like that of all living things, possesses a fundamental passive integrity that in other respects allows them to be active. However, let us focus on the active aspect of organisms.

“I don’t know what I’m doing!” “No, neither do I!”
© istockphoto.com/Antrey

What is the source or director of activity in living things that are automatic in their behavior, as ants are? This is not a great mystery, even if we do not know the finest-grained details: animal behaviors derive from a delicate interplay between biological structure, inherited instinct, acquired learning, and environmental input. The behavior is automatic rather than chosen, but that is not to say that it is deterministic or inevitably any given response, since so many of the factors, and certainly their interactions, will be unpredictable in their results.

Notice however that there is no serious question about the appropriateness of automatic animal behaviors. The animal behaves as it does, it cannot help it, and it makes no sense to ask whether what it did was ‘right’. We might judge an animal’s action against the statistically normal activity of its species in order to say that it is doing better or worse from a natural selection point of view, or we could imagine a version of the same animal with better instincts, but it is not as if the malfunctioning or inferior animal should have known better or should have thought more carefully. We do not hold its automatic behaviors against the animal. Yet we do make such judgments about knowledge and rationality – which goes to show that none of this automatic behavior concerns epistemology yet. To make knowledge possible, another level of complexity is required.

Representation-Governed Behavior

A great deal of the behavior of all animals, including of human beings, is governed automatically. But some animals have more flexible capacities than what purely automatic living things can exploit. Some animals possess minds that enable them actively to seek equilibrium with the environment in a way that is different from both rocks and ants. Having a mind entails the capacity to represent other things, and this implies a rich repertoire of interacting, interrelated representations.

Mental representations are a kind of mental intermediary ‘between’ a living thing and the world to which it responds (‘between’ must be in quotation marks because, of course, the living thing carries this intermediary within itself). One way to put it is that cognitive beings behave on the basis of mental ‘descriptions’ of the world that say something to the cognitive agent about how the world is. Descriptions can be language-like, as they may be for human beings; or they can be given in terms of some other medium that describes, such as in images. A useful metaphor here is of a cognitive agent’s representations as comprising a map. We can imagine that the cognitive agent carries around a small book of such maps, and that it is the maps that she is responding to mentally. These maps describe the environment using resources that are quite different from the environment itself. Nor does the map have to be entirely accurate to enable successful navigation. In many cases complete fidelity would be cumbersome and distracting. Rather, a useful map needs to encode a description of the environment that suits the immediate needs and purposes of the map user. Thus representations might be something very simple like the orientation of a line on a two-dimensional surface, or might be very complex like the conceptual description of a social situation. One important branch of psychology, cognitive psychology, investigates the representations had by human beings, and by some non-human animals.

Instead of being wholly governed by instinct and conditioning, the descriptions cognitive beings carry around can interact in complex ways (eg, in thinking). A single representation might be triggered by very diverse stimuli, or trigger different responses, or diverse representations might well all trigger the same response. So the representations can interact independently of the world to some degree, in the sense that they are not locked into single pairs of stimulus/response relations. So descriptions are immensely useful in that, if the interaction of these intermediaries is governed by rules different from the rules governing what is happening in the environment, the agent has a rich set of extra resources to exploit which afford a flexibility in behavior that no merely mechanically active agent has. Therefore the point of decoupling behavior from stimuli through representations looks to be to give active beings a chance to operate in an environment distinct from the physical environment. We cognitive beings ‘behave’ in a world that does not physically exist, the world of the descriptions, in order to consider and make plans about situations that do not now exist. Here the stakes are much lower and we can go back to earlier steps in our plans to try out different behaviors. That is what planning is.

Neither right nor wrong: white blood cells zealously guarding your veins

The most important consequence for epistemology of under-standing cognitive beings in terms of representations, is that the way they represent may at times be inadequate for the purpose of staying in equilibrium with the environment, or just plain wrong. With merely automatic behaviors, it did not make sense to ask whether an action is right or wrong. If a white blood cell attacks healthy tissue, we would not say that the cell “thought the tissue was an antigen, but was wrong.” Well, we might say this, but we would be talking wholly metaphorically. That is because cells do not think anything, they just automatically behave. But in the case of a human being who mistakenly grabs a piece of plastic fruit from a bowl when she is hungry, it is not in a metaphorical but in a literal sense that we say “she thought it was a real piece of fruit, but was wrong.” The way we can make sense of the difference is by claiming that human beings, but not cells, represent the world. In this case the human being is incorrectly representing the objects before her as real fruit. The description that is governing her behavior is wrong – although it will be quickly corrected.

Epistemology As About Correctness

We may now summarize the big picture of epistemology as follows. Non-living things can be connected and arranged into living things in distinctive ways, to achieve persistence and integrity in the environment in a manner that cannot be achieved by non-living things individually – when these non-living parts comprise an active organism. When living things reach a certain level of complexity, one way they may increase in survival fitness is for their behaviors to be decoupled from automatic responses to the environment. The way to break this direct connection is for there to be an intermediary between the stimulus and the response, so that the organism’s response is to the intermediary rather than to the stimulus. We can then legitimately ask whether the intermediary is an adequate representation of the environment. It may not be, and the downside, when the representations are flawed, may be considerable. But the upside of having correct representations is enormous. By and large, evolution has made sure that in normal cases the representational intermediaries are not too far out of step with what is required for a cognitive animal to be successful in their particular environment.

We are already well toward our goal of locating epistemology in a broader understanding of the function of mind. Earlier, we identified good reasons as ones governed by rules that keep beliefs grounded in stable sources of evidence about the world, such as perception and (at least in some cases) memories and testimony. We can now say that the rules of good reasoning range over representations. In fact, the capacity for representations is essential for there to be epistemological rules at all.

For some epistemologists, investigating the unconscious (automatic) rules of thought is the most important part of epistemology. But a final step to epistemology in the broadest sense has to do with consciousness. When representations are available in the form of language, we can even string them together into articulate descriptions regarding everything from the most mundane aspects of life to the most intricate scientific theories. When we form such representations with the goal of avoiding error, they comprise our beliefs. Moreover, when representations turn into beliefs, we can investigate and question them explicitly by asking what makes some beliefs good and others bad. Relatedly, we can also ask, how can a self-conscious agent maximize her good beliefs? Are there habits, procedures, or mental policies that help us generate beliefs that are more adequate to or more accurately representative of the world? These are the sorts of questions traditionally asked by epistemologists.

The freedom, flexibility and adaptive advantage of the capacity for belief also comes with costs. Uncountably many things can influence the way we represent the world, and many of them can draw our beliefs away from a correct description of the world. Epistemology, then, is about closing the gap between the beliefs we have and the best beliefs we can obtain. It is the study of how and what it means to have representations that are well-enough matched up to situations that those beliefs will yield successful behaviors. The most successful beliefs, that is, the ones that show themselves to be such that they can perpetually keep us in organic equilibrium with the environment, are what we call ‘knowledge’. The beliefs that are more provisional, but which are related to the other beliefs in such a way as to give them considerable pedigree in keeping us in equilibrium, we may call ‘rational.’ Thus we have arrived back at epistemology, but by way of trying to understand natural sophisticated active living cognitive beings in the world.

© Dr Joseph L. Cruz 2013

Joe Cruz is a professor of philosophy and chair of the cognitive science program at Williams College in Williamstown, Massachusetts. He is the co-author, with John Pollock, of Contemporary Theories of Knowledge, 2nd Edition (Rowman & Littlefield, 1999).

• Thanks to Margaret Coady, Kate Nolfi, and numerous students for discussion on these topics. And, as always, thanks to the late John Pollock.

This site uses cookies to recognize users and allow us to analyse site usage. By continuing to browse the site with cookies enabled in your browser, you consent to the use of cookies in accordance with our privacy policy. X