On Philosophy

September 1, 2006

August’s Top 5

Filed under: General Philosophy — Peter @ 7:30 pm

First and Third Person Approaches to Knowledge
On Knowledge
What is Truth?
The Intellectual Bankruptcy of Conceptual Analysis


Essay: Human Knowledge, Animal Knowledge, and the Role of Reflective Justification

Filed under: Epistemology,Essays — Peter @ 12:01 am

Some philosophers, disagreeing with Kornblith, think that there is something special about human knowledge that sets it apart from animal knowledge. One way to make this distinction is by arguing that human knowledge is justified in a way that animal knowledge is not. One could claim that human knowledge only exists when the person who has that knowledge has reflected on it and constructed reasons for it. Of course the most common way to construct such reasons would be through introspection, the process of looking inwards and discovering what reasons you actually have for believing it. This reliance on introspection was initiated by Descartes, who thought that with reflection we could examine each of our beliefs and know the reasons that support it, and this idea has found its way into many theories that would separate human knowledge from animal knowledge since then. To many modern philosophers, such as Chisholm, human knowledge is founded on a process of inner questioning and justification, which would seem to rule out the possibility of animals having knowledge.

Kornblith is, of course, against such a definition of knowledge. He argues that knowledge cannot be based on introspection because introspection is well known to be unreliable. For example we know that people have unconscious biases towards selecting items on the right, and that they are not aware of these biases [1]. Even when people are informed that such biases exist they will deny that they are influenced by them. So, not only does introspection fail to reveal some of the reasons for why we make choices, but it is unable to reveal to us that it is incomplete. However there are more problems with introspection than just this. Even if we are able to detect in some cases when introspection is in error we will be unable to correct for these mistakes, because we will mistakenly put more emphasis on our successes at introspection than on our failures, a phenomena called confirmation bias. Thus not only will introspection be in error, but it can’t be relied on for improvements in the way we approach introspection or finding reasons either [2]. Because of these problems Kornblith concludes that introspection is an unsuitable foundation for knowledge.

Kornblith also has two additional arguments against approaches that rely on some kind of justification inaccessible to animals, one against a coherence account and one against a foundationalist account. A coherence account can’t describe real knowledge, according to Kornblith, because requiring new knowledge to cohere with existing knowledge is impossible. The amount of knowledge we already possess is extremely large, and to ensure that a new statement was coherent we would have to compare it against all of our previous knowledge, at once (in case some combination of established knowledge was incoherent with it). But, since we possess limited mental faculties, this task is impossible [3]. Similarly foundationalism also requires an impossible task, evaluating a belief in light of the total evidence. The volume of evidence for many beliefs can be so great that it would be impossible to hold it in mind at once, as required by our desire to evaluate our new belief [4].

Kornblith has successfully argued against a distinction between human and animal knowledge based on introspection. However, he has not ruled out the possibility that reflective justification, not based on introspection, distinguishes human knowledge from animal knowledge. In fact Kornblith’s demonstration of the fallibility of intuitive judgments seems to make justification all the more essential to knowledge. Let us return to the example of picking items on a shelf. Let us further assume that the item on the right is usually not a bad choice, so in some sense the belief may even be reliable, and possibly knowledge according to Kornblith. The person making the choice, if they attempt to justify their actions, will surely come up with reasons for making that decision. However justification is not simply coming up with whatever reasons we feel like; that would be just as bad as relying on introspection. These reasons are instead reflected on, and compared to reasons for selecting other choices instead. It is true that there will still be an unconscious bias to favor the reasons in favor of the rightmost item. Even so this justification process makes it more likely that we will overcome our unconscious biases and make a better choice. We might even view, in a naturalistic sense, this kind of reflective justification as “super reliable”, more reliable than the non-reflective processes by which animals come to make decisions [5]. We then could argue that this new natural kind is real knowledge, and that the natural kind that Kornblith has identified is simply proto-knowledge.

Even if this account seems reasonable we still require a response to Kornblith’s argument against foundationalism, since it is typical to think of justification as requiring a foundation to be rational. Of course we could take a “naturalistic” approach to justification, and argue that it is simply a tool for making our choices more reliable, and thus that it doesn’t require any absolute foundation, since we aren’t arguing that it has normative powers. Or we could argue, again naturalistically, that justification doesn’t require support by all the available evidence, only enough to make it more reliable than intuition. But these are not the kind of arguments I would make. I think that it is possible to defend justification, in the form of a hypothesis supported by all the available evidence. A statement is supported by all the evidence if and only if there is no evidence that contradicts it. Given this a person attempting to justify a hypothesis could scan their memory looking for contradictory cases, and if they find none they can conclude that the hypothesis is supported by the evidence, and since they have to examine the evidence only one piece at a time, and since the unconscious is very good at retrieving relevant information, this can be done with limited mental resources. Of course there is still the problem of confirmation bias, which says that we are likely to ignore the evidence against a hypothesis we favor. All this proves, however, is that people are fallible when they justify, which should be no surprise. Still, we can learn to overcome our confirmation bias and approach the ideal with effort, and the mere fact that the ideal exists and is approachable is enough to defend justification as a normative basis for knowledge. Thus it seems that there is a natural kind of belief that only humans, as sufficiently intelligent animals, possess, a kind that seems best described as knowledge, meaning that what we thought was knowledge in animals must really have been something else.


1. “… we should not be influenced by their relative position, nevertheless, it seems that we are. Subjects in this study, however, are unaware that their judgments are affected by the relative position of the goods being evaluated. Moreover … asking there subjects to introspect more carefully … would not make them better acquainted with the source of their judgments.” Kornblith, Knowledge and its Place in Nature, page 111

2. “The existence of confirmation bias provides a global reason for thinking that responsible agents are unlikely to discover, at least by way of introspection, the extent of their epistemic shortcomings. Most agents thinks of themselves as tolerably reliable in their acquisition of belief … our natural confirmation bias will serve to insulate it from disconfirming evidence.” Kornblith, Knowledge and its Place in Nature, page 118

3. “It is not simply that human beings have difficulty in determining the consistency of large sets of sentences. It is simply beyond the powers of any possible computational device… In so far as coherence requires consistency we can be sure our beliefs are not the product of mechanisms that determine coherence because such mechanisms are fundamentally impossible.” Kornblith, Knowledge and its Place in Nature, page 129

4. “The appeal to the agent’s total evidence is absolutely essential here, because the probability of a proposition cannot be determined by appeal to anything short of the totality of the agent
s evidence. But … the total evidence requirement would thereby require a reflexive grasp of the agent’s total evidence, something very far out of the reach of any agent.” Kornblith, Knowledge and its Place in Nature, page 134

5. This is not to say that only humans could reason in this way, simply that it is a function of intelligence, and it does not seem that any other animals have this capacity (although they might, you never know).

Blog at WordPress.com.