What is Human Knowledge? Categorical Normativity and Categorical Epistemology, #1.

Grand Central Station, New York City (circa 1930)

The essay below will be published in two installments; this is the first.

You can also download and read or share a .pdf of the complete text of this essay, including the list of REFERENCES, by scrolling down to the bottom of this post and clicking on the Download tab.

What is Human Knowledge? Categorical Normativity and Categorical Epistemology, #1.

By a veridical appearance I mean anything X that appears as F, or appears F-ly, or appears to be F, to any or all rational human cognizers, just insofar as, and precisely because, X really and truly is F. For example, if I say “It appears that sunlight is cascading into Grand Central Station on some day circa 1930,” as per the above classic photograph, or “It appears that someone attempted to assassinate Donald Trump on 13 July 2024,” or “It appears that 2 + 2 = 4,” or “It appears that The Minimal Law of Non-Contradiction—i.e., that not every statement is both true and false (Putnam, 1983; Hanna, 2006)—applies universally,” and what I say is literally correct, i.e., literally the way the manifestly real world is, then all the things I am talking about are veridical appearances. By the manifestly real world, I mean the world as it veridically appears to any or all rational human cognizers or agents. And finally, by truth I mean that a proposition, statement, or indicative sentence is true if and only if what it says is manifestly real.

Moreover, I believe that the primitive fact of what I call categorical normativity inheres in all rational human intentionality whatsoever—including all rational human consciousness, mental content, belief, and knowledge. What is this primitive fact? Normativity, as I am understanding it, consists in the fact that all minded animals, whether merely sentient (conscious minimal agents) or also sapient (self-consciously conscious, rational agents), have desires, aims, commitments, ends, goals, ideals, and/or values. Now insofar as sapient or rational minded animals naturally treat these aims, commitments, ends, goals, ideals, and/or values as rules or principles for guiding theoretical inquiry and practical enterprises, as reasons for justifying beliefs and intentional actions, and also as standards for critical evaluation and judgment, then at least some of those rules, principles, reasons, and standards are non-instrumental, unconditional, desired for their own sake as an end-in-themselves, non-pragmatic, non-prudential, and obtain no-matter-what-the-consequences. These are categorical norms, and my claim is that they necessarily inhere in all rational human caring. Categorical norms are perfectly consistent with norms that are instrumental, conditional, desired for the sake of other ends, pragmatic, prudential, or obtain only in virtue of good consequences. Nevertheless, categorical norms are necessarily underdetermined by all other sorts of norms—that is, categorical norms do not strongly supervene on any other sorts of norms—and therefore they cannot be assimilated to or replaced by those other sorts of norms. Correspondingly, categorical norms provide overriding reasons for belief and intentional action.

In order to make my next point, I’ll have to pause briefly for some slightly technical terminology and definitions. Strong supervenience is a strict determination-relation between sets of properties of different ontological “levels,” a relation that is weaker than strict property-identity, and is usually taken to be asymmetric, although two-way or bilateral strong supervenience is also possible. But assuming for the purposes of simpler exposition that strong supervenience is asymmetric, then, more precisely, B-properties (= the higher level properties) strongly supervene on A-properties (= the lower-level properties) if and only if (i) for any property F among the A-properties possessed by something X, F necessitates X’s also having property G among the B-properties (upwards necessitation), and (ii) there cannot be a change in any of X’s B-properties without a corresponding change in X’s A-properties (necessary co-variation) (see, e.g., Kim, 1993: esp. part 1; Chalmers, 1996: chs. 2-3; and Horgan, 1993).

It follows from strong supervenience that any two things X and Y share all their A-properties in common only if they share all their B-properties in common (indiscriminability). For example, if heat strongly supervenes on mean molecular motion, then two things have the same kinetic molecular properties in common only if they have the same temperature properties in common. Facts are just actual or possible instantiations of properties. Hence strong supervenience for properties entails strong supervenience for facts, and failures of strong supervenience for properties correspondingly entails failures of strong supervenience for facts.

Now logical strong supervenience is strong supervenience that also obtains with logical, analytic, or conceptual a priori necessity. The strict “downwards identity” of higher-level properties with corresponding lower-level properties entails logical strong supervenience; but logical strong supervenience is also consistent with the multiple instantiability or realizability of the same higher-level properties across different lower-level properties, hence consistent with “downwards non-identity.” For example, even if human body heat logically strongly supervenes on mean molecular motion inside the human body, it remains conceivable and possible that the very same temperature properties are instantiated or realized in humanoids made out of quite different kinds of stuff that nevertheless plays the same mean-molecular-motion role—for example, the “replicants” imagined in Philip K. Dick’s classic science fiction novel, Do Androids Dream of Electric Sheep? and in Ridley Scott’s equally classic sci-fi movie, Blade Runner (Dick, 1968; Scott, 1982). Hence logical strong supervenience is the most inclusive reductive metaphysical relation.

Here’s the point I want to make by using the notion of logical strong supervenience. If a norm really is categorical, then it cannot be reduced to contingent physical facts or natural causal laws by means of logical strong supervenience. This is shown by the following reductio argument.

1. Suppose that categorical norms are reducible to contingent physical facts or natural causal laws.

2. Contingent physical facts and natural causal laws are inherently conditioned by, and conditional upon, the actual spatiotemporal locations of those facts and the actual constitution and distribution of matter and forces in the physical world, whereas categorical norms are inherently unconditioned and unconditional.

3. But the explanatory reduction of X to Y entails showing that X is, at the very least, logically strongly supervenient on Y,

4. So by the initial supposition made in 1., categorical norms would then be logically strongly supervenient on inherently conditioned, conditional facts.

5. But then categorical norms are both inherently unconditioned and unconditional and also strictly dependent on what is inherently conditioned and conditional, which is a contradiction.

6. Therefore, categorical norms cannot be reduced to contingent physical facts or natural causal laws, by reductio ad absurdum as applied to the initial supposition made in 1.[i]

In what follows, by a conscious-evidence-based reason, I mean a reason that is based on evidence provided by a conscious act, state, or process. And by a conscious act, state, or process I mean a subjectively-experienced, intentionally-directed mental act, state, or process. Thus reasons that are based on our capacities for sense perception, memory, imagination, apperception or self-consciousness, judgment (including the reception of testimony), deductive inference, inductive inference, abductive inference, mathematical intuition, logical intuition, or philosophical intuition, are all conscious-evidence-based reasons.

My account of the nature of knowledge, categorical epistemology, is robustly normative in character, and it also flows naturally from the widely-known and almost universally-accepted “Gettier counterexamples” to the classical analysis of knowledge, according to which knowledge is the same as justified true belief (Gettier, 1963; Shope, 1983; Ichikawa and Steup, 2024).

Duncan Pritchard and others have correctly pointed out that the Gettier cases show that the classical analysis of knowledge leaves justified true belief open to luck, that is, to merely accidental or contingent connections between justifying evidence and the truth-makers of beliefs. Hence, in addition to justified true belief, authentic knowledge further requires the satisfaction of an anti-luck, or externalist,condition. Pritchard and others have also correctly pointed out that the classical analyis of knowledge fails to require that cognitive subjects acquire their justifying evidence via properly-functioning cognitive capacities or mechanisms. Hence authentic knowledge also requires the satisfaction of a cognitive virtues, or virtue epistemology, condition (Pritchard, 2012). What I will call High-Bar knowledge includes maximally strong versions of both the anti-luck condition and the cognitive virtues condition alike, as well as requiring the satisfaction of an evidential-phenomenological, or internalist, condition, and in this way it also rules out global or radical skepticism.

Here’s what I mean by all that. The simplest kind of Gettier counterexample goes like this. I look at my iPhone, and it says that it is 7:00 am. And I know by experience that my iPhone has been working fine for months. So I have a conscious-evidence-based reason for asserting that it is 7:00 am. And, as it happens, it manifestly really is 7:00 am, and the statement “It is 7:00 am” is true. But, unbeknownst to me, my iPhone has been broken since 7:00 pm last evening, when, by a malfunction of the digital mechanism, it started reading 7:00 am and froze at that setting; and I have not looked at it since then. So even though I have a conscious-evidence-based reason for asserting that it is 7:00 am, and even though it it is true that it is 7:00 am, and I believe that it is 7:00 am, I do not know that it is 7:00 am. So, it would seem that knowledge is not justified true belief.

How should we understand this result? My own take on the Gettier counterexamples is that although knowledge really is justified true belief, the counterexamples initially suggest the opposite, by trading on a special internal normative feature of the concepts and facts of epistemic justification and knowledge: epistemic justification and knowledge are normatively two-dimensional, in the sense that by their very nature they are either Low-Bar or High-Bar. Let me now, in turn, explain what I mean by this.

Low-Bar. The “Low-Bar” dimension of epistemic justification allows for justification to be more or less detached from truth, and means: “whatever provides a conscious-evidence-based reason for the believer to assert her belief-claim, even if that belief turns out false,” in which case that belief obviously is not knowledge in the normatively highest sense. But most importantly for the Gettier counterexamples, what I will call Low-Bar justification is also consistent with cases (like the broken iPhone case) in which the believer’s claim is actually true, yet that actual truth is neither inherently or intrinsically connected to the believer’s conscious-evidence-based reason for asserting her belief-claim, nor even in a context-sensitive way, causally reliably connected to the believer’s conscious-evidence-based reason for asserting her belief-claim. Otherwise put, the truth of the claim in these cases is only accidentally or contingently connected to the believer’s conscious-evidence-based reason for asserting her belief-claim. That is Low-Bar justification.

Now this clearly and distinctly points up the fact that knowledge in the normatively highest sense, or what I will call High-Bar knowledge, requires an inherent or intrinsic connection—i.e., a non-accidental or necessary connection—between the truth of a believer’s belief-claim and a believer’s sufficient conscious-evidence-based reason for asserting her belief-claim, as delivered by her properly-functioning cognitive capacities or mechanisms. That is, it requires what I call High-Bar justified true belief. This is because in the cases in which there is only an accidental or contingent connection, the believer’s belief-claim could just as easily have been false with no change whatsoever in the believer’s conscious-evidence-based reason for asserting her belief-claim. So knowledge in the normatively highest sense, that is, High-Bar justified true belief, is not the same as Low-Bar knowledge, which involves justified true belief in the Low-Bar sense only. In that sense, High-Bar knowledge is not Low-Bar justified true belief, although High-Bar knowledge still is and always will be High-Bar justified true belief. Correspondingly, Low-Bar knowledge still is and always will be Low-Bar justified true belief. Hence, provided that we keep our bar-levels straight, knowledge really is justified true belief.

High-Bar. By sharp contrast, then, the “High-Bar” dimension of knowledge and justification requires that belief be inherently or intrinsically connected to truth, via the properly-functioning cognitive capacities or mechanisms of the cognitive subject, and means: “whatever provides a sufficient conscious-evidence-based reason for the believer to assert her belief-claim, via her properly-functioning cognitive capacities or mechanisms, and also is inherently or intrinsically connected to the truth of that belief-claim.” Otherwise put, High-Bar knowledge has the following three fundamental features. First, belief is self-evident, i.e., intrinsically compelling, thereby satisfying an evidential-phenomenological or internalist condition on authentic knowledge. Second, this self-evidence is informationally delivered to belief by a properly-functioning cognitive capacity or mechanism, thereby satisfying a cognitive virtues condition on authentic knowledge. And third, belief provides a non-accidental or necessary tie to the truth-makers of belief, thereby satisfying an anti-luck or externalist condition on authentic knowledge.

An example of High-Bar knowledge would be a case that is radically different from any sort of Gettier case, and also radically different from any other sort of “bad” epistemic case involving falsity or failed justification. In this all-around good epistemic case, as a paradigm, and indefinitely many others relevantly like it, I objectively know, via basic authoritative a priori objectively necessarily true mathematical rational intuition, that

3+4=7, i.e., | | | + | | | | = | | | | | | |

and thereby achieve High Bar a priori knowledge. Now by an essentially reliable cognitive capacity or mechanism, I mean a cognitive capacity or mechanism that tracks truth counterfactually and in a context-sensitive way across all relevantly similar metaphysically possible worlds. So High-Bar justified true belief is the same as High-Bar knowledge, precisely because justification occurs by means of an essentially reliable cognitive capacity or mechanism, in this case, basic authoritative mathematical rational intuition.

This paradigmatically good epistemic case should also be distinguished from another variant case in which my iPhone says it is 7:00 am, and my iPhone is still working fine, and it is actually 7:00 am, I believe that it is 7:00 am, the statement “It is 7:00 am” is true, and it is also the case that (i) whenever, in relevantly similar cases, it were to be such-and-such a time, call it T, and I looked at my my iPhone and it read “T,” then I would believe that it is T, and (ii) whenever, in relevantly similar cases, it were, by some salient difference, not to be T and I looked at my iPhone, yet my iPhone still read “T,” then I would not believe that it is T and would instead believe that my iPhone was malfunctioning. So I know that it is 7:00 am, because my conscious evidence for asserting my belief is connected to the truth of that belief-claim with context-sensitive causal reliability. Now by a context-sensitive causally reliable cognitive capacity or mechanism I mean a cognitive capacity or mechanism that tracks truth in the actual world, and also counterfactually and in a context-sensitive way across all relevantly similar nomologically possible worlds. In this “pretty good” case, then, the context-sensitive causally reliable cognitive capacity or mechanism is my capacity for veridical, direct sense perception,together with a further online—that is, currently properly functioning—capacity of mine for detecting salient breakdowns of my iPhone whenever they occur.

But context-sensitive causally reliable knowledge, “pretty good” though it is, is not the normatively best or highest kind of knowledge, precisely because the connection between my conscious-evidence-based reason and the truth-maker of my belief is not inherent or intrinsic. On the one hand, context-sensitive causally reliable knowledge is open to global skeptical worries: in at least some introspectively indistinguishable, conceivably possible worlds containing the very same conscious-evidence-based reason, that belief is instead connected to a falsity-maker, not a truth-maker (see, e.g., Cohen, 1984). This is also called “new evil demon” skepticism, to distinguish it from classical Cartesian evil demon skepticism, which of course postulates the conceivable possibility of falsity-makers in the actual world for any and all seemingly true beliefs.

And on the other hand, even given context-sensitive causally reliable knowledge, it is not as if my capacity for veridical, direct sense perception, together with my capacity to detect salient iPhone breakdowns, completely convincingly, intrinsically compellingly, or self-evidently “locks onto” the context-sensitive causal sequence that ties my well-functioning iPhone to the US standard atomic clock (or whatever) that grounds it. That is, even given context-sensitive causally reliable knowledge, it is not as if I have rational insight into the underlying structure of what connects my conscious-evidence-based reason for believing to the truthmaker of my belief. Indeed, my conscious-evidence-based reason for believing could be epistemically flawed in various ways, including greater or lesser irrelevance to the situation at hand, greater or lesser superficiality, greater or lesser triviality, or more or less obvious formal inconsistency with other beliefs I hold.

This point is also brought out clearly, although in a sense unintentionally, by Keith Lehrer’s well-known “Truetemp” thought-experiment, whose explicit aim is to show that context-sensitive causally reliable true belief is not the same as knowledge (Lehrer, 1990: pp. 163-164). Lehrer’s example features a context-sensitive causally reliable temperature-reading device connected (unbeknownst to Mr Truetemp himself) to Mr Truetemp’s brain, that together with his brain yields a context-sensitive causally reliable cognitive capacity or mechanism for his beliefs about temperature. This example, in turn, is supposed to trigger our judgment that Mr Truetemp’s context-sensitive causally reliable true beliefs about temperature are not knowledge. But in fact, what the Truetemp case shows, just like my iPhone case, is simply that context-sensitive, causally reliable Low-Bar knowledge, even though it is pretty good, is not the same as High-Bar knowledge. Otherwise put, my context-sensitive causally reliable perceptual knowledge that it is 7:00am by looking at my iPhone is not essentially reliable, as it is in the paradigmatically good epistemic case where I know that

3+4=7, i.e., | | | + | | | | = | | | | | | |

via basic authoritative mathematical rational intuition.

In this way, what the Gettier counterexamples and their variant cases show us are four distinct synthetic a priori philosophical truths about knowledge. First, High-Bar knowledge is not the same as Low-Bar knowledge, that is, High-Bar knowledge is not the same as Low-Bar justified true belief. Second, High-Bar knowledge is also not the same as context-sensitive causally reliable Low-Bar knowledge, that is, High-Bar knowledge is not the same as context-sensitive causally reliable Low-Bar justified true belief, which in turn is distinct from mere Low-Bar knowledge, or Low-Bar justified true belief. Third, High-Bar knowledge isthe same as High-Bar justified true belief, or essentially reliable justified true belief. Fourth and finally, Low-Bar knowledge is the same as Low-Bar justified true belief; context-sensitive causally reliable Low-Bar knowledge is the same as context-sensitive causally reliable true belief; and High-Bar knowledge is the same as High-Bar justified true belief. Therefore, provided we keep our bar-levels straight, knowledge really is justified true belief.

The leading notion of categorical epistemology is what I am calling High-Bar knowledge. Any theory of knowledge that adequately establishes an inherent or intrinsic connection between the sufficient conscious-evidence-based reason for a believer’s assertion of her belief-claim, via her properly-functioning cognitive capacities or mechanisms, and the truth of her belief, also shows that this is an essentially reliable belief. This theory thereby constitutes an adequate philosophical explanation of the highest kind of knowledge, which in turn counts as the highest good, or summum bonum, of epistemology. And that is High-Bar knowledge.

NOTE

[i] For an analogous argument against the very idea of scientific naturalism and reductive physicalism as applied to logic (aka, logical psychologism), see (Hanna, 2006: ch. 1).

Download