Unlike Quine, Murray Clarke is defending a weak replacement thesis of naturalized epistemology. Clarke argues that accurate indexical representations (a belief about the here and now which moves us to action) have been crucial for the survival and reproduction of homo sapiens. He suggests that reliable processes have been “selected for” because of their indirect, but close, connection to true belief during the hunter-gatherer period of our ancestral history. True beliefs are not heritable, reliable processes are heritable. Those reliable processes connected with reasoning take the form of Darwinian Algorithms: a plethora of specialized, domain-specific inference rules designed to solve specific, recurrent, adaptive problems in social exchange contexts. Humans do not reason logically, but adaptively. So, the mind is composed of Darwinian modules (a computational processor) and Chomskian modules (sets of representation or data). Darwinian module (DM) is represented as hardware and Chomskian Module as software.

Clarke’s account of knowledge

Based on the notion of Darwinian modules, Clarke’s account of knowledge is as follows:

  • P is true,
  • S believes that P
  • S’s belief that p is DM-justified.

< S’s belief that P is DM-justified at t > iff

  • P is caused by a Darwinian module that has been instantiated.
  • There is no other reliable or conditionally reliable Darwinian module available to S that, had it been used by S in addition to the module actually used, would have resulted in S’s not believing that P at t.

Clarke vs Stich

Clarke claims that a generally false belief set cannot be an advantage for humans. If I am to eat that red berry on that shrub, it had better be the case that its consumption will not terminate my life. Though there may be exceptions, cases where “ignorance is bliss”. If I believe that red berry is not poisonous when it is poisonous then I am a dead (false negative). If, on the other hand, I believe that red berry is poisonous when it is not, then I pay no penalty (false positive).

Stephen Stich argues that adopting a cautious, risk-aversive inferential strategy that results in mostly false positive beliefs may well tend to produce a higher level of external fitness than adopting a risky inferential strategy that results in mostly false negative beliefs. External fitness is such that one genetic program is fitter than another if its input/output pairings are more conducive to survival and successful reproduction. Clarke denies the Stitch’s argument saying that particular false beliefs can be very bad for your health (having bad visual perception during a road trip). Generally, false beliefs will not be more externally fitness enhancing than true beliefs.

Clarke’s argument about reliable processes

(1) It is physically likely that processing mostly true beliefs would be an evolutionary advantage.
(2) But evolution cannot select for true belief directly.
(3) Reliable processes could produce mostly true beliefs and such processes could be selected for.
(4) There are no alternative mechanisms that are likely to have produced mostly true beliefs.
∴ Therefore, reliable processes may well have been selected for.

Selection for:
To say that there is a selection for a given property means that having that property causes success in survival and reproduction. E.g. Jaw structure properties for mastification (chewing). ‘Selection for’ describes causes of the selection process.

Selection of:
To say that a given sort of object was selected is merely to say that the result of the selection process was to increase the representation of that kind of object. E.g. the human chin. ‘Selection of’ describes effects of the selection process. Objects that are selected of are termed “free riders” because they need not confer any evolutionary benefit in terms of survival and reproduction. The sort of abstract reasoning involved in science, mathematics, and logic is the result of a free rider effect.

Clarke claims that there are no universally reliable reasoning processes, but only “reliable-in-a-social-exchange-content” inference rules. Psychologists found that how humans reason depends on the nature of the subject matter that they are asked to reason about. We reason much better with cost benefit structures (social contracts). For example, being right is nice, but being right when there are benefits or costs at stake is “a biological imperative”.

Clarke vs Goldman

The reason we want justified beliefs is that such beliefs have a good chance of being true. When we know something it is not an accident. Goldman’s reliabilism provides a linkage between justification and truth, but it is flawed in several cases, such as: the lottery paradox and the generality problem.

The lottery paradox
Suppose there is a reliable belief-forming mechanism available to S such that S believes by appeal to reason that in a fair lottery the chances that S will lose when she buys a ticket are at least 0.99. Such a belief would be applied for each ticket holder, yet someone must win. Hence, Goldman’s account gets the wrong result here.

The generality problem
The reliable process account does not specify the degree of generality of process types. For instance, beliefs about cows arrived at from ten feet away seems to be far more highly justified than beliefs about cows arrived at from half a mile away.

Goldman accepts the evolutionary connection between justification and truth but fails to explain the exact nature of the evolutionary connection that makes this possible. His account fails to explain how knowledge is possible.

Clarke vs Dretske

Dretske is an internalist and he denies that justification is related to knowledge. Dretske and other internalists rely on reason and evidence for our beliefs. However, there is no sort of guarantee that possessing a reason gives a true belief.

So, neither Goldman nor Dretske provides an adequate account of justification and an adequate account of knowledge. This is because neither account succeeds in showing the evolutionary connection between justification and truth that leads to knowledge.

Reconstructing knowledge

Non-meliorative notion concerns already formed beliefs and we think what objections can be applied to them. Example: traditional account of knowledge.
Meliorative notion concerns proper scientific methods: how we should pursue inquiry, what sort of methodology can be applied in that pursuit.

Carnap was concerned with meliorative project, while Goldman talks about both non-meliorative and meliorative projects. To answer the meliorative question, we should answer the non-meliorative question. Science (Carnap’s account) wanted to change the subject. Internalists and Externalists did not modified it, but provided the main precondition of knowledge.

Goldman doesn’t pursue an inquiry, he simply gives truth conditions. Carnap lays out an account of inductive reasoning: how human becomes a rational agent. Both notions play a crucial role in the reconstruction of the foundations of our knowledge of the external world. Suppose that we should determine the medicinal value of D and the means requires that one use the double-blind testing methodology (claim). Carnap’s inductive logic might be used to determine how much epistemic support the claim has relative to the evidence. What Carnap does not provide is an account of non-meliorative justification that would ground his inductive logic. Goldman’s account could be used to underwrite Carnap’s meliorative account of justification. In that sense, Carnap’s inductive logic along with Coldman’s reliabilism constitute good reasoning.

Meta-EpistemicsGoldman’s account (the meaning of justification)
Normative EpistemicsCarnap’s inductive logic (specific principles)
Applied EpistemicsInstrumental rationality
Drug test

Hence, any adequate epistemology must provide a metaepistemic account in order to ground its “normative epistemic” account of the meliorative project. Subsequently, the account of instrumental rationality would be entailed by one’s meta-and normative epistemic theory. A complete epistemological theory would thus be tripatite: consisting of meta-, normative and an applied epistemic account of its subject matter.

Clarke vs Nonnaturalism

As a naturalist, Clarke agrees that science must play an important role in clarifying normative, evaluative epistemic terms in nonnormative, descriptive terms. He also thinks that a properly naturalized epistemology should use the latest results from empirical psychology.

According to naturalists, knowledge is a real phenomenon in the world. It is a natural kind. It is not some sort of concept that we created.

The “Knowledge is a natural kind” argument

  1. Either knowledge exists or knowledge does not exist.
  2. If knowledge exists, it is either a conceptual kind or a natural one.
  3. If knowledge does not exist, the scepticism is true.
  4. Skepticism is false (at least most analytics believe that scepticism is false).
  5. Knowledge exists. (3,4)
  6. Knowledge is a conceptual kind or a natural kind.
  7. If knowledge is a conceptual kind, then it is a social construct.
  8. If knowledge is a social construct, then epistemic relativism is true.
  9. Epistemic relativism is false.
  10. Knowledge is not a social construct. (8,9)
  11. Knowledge is not a conceptual kind. (7,10)
  12. Therefore, knowledge is a natural kind. (6,11)

Suppose water is a natural kind. As a natural kind H2O includes:

  1. Homeostatic properties (specific, stable structure at the microlevel)
  2. Microproperties (colorless, tasteless)
  3. Causal laws (when we heat it, it boils)

So, water is something that is discovered in the world, has stable configuration and supports causal laws. If the claim that “knowledge is a natural kind” is true, then it should constitute of (1), (2), (3).

Bibliography:

  • Clarke, Murray, Reconstructing Reason and Representation, Bradford Books, The MIT Press, 2004