Keith Lehrer on Self-Trustworthiness

Len O’Brian
Keith Lehrer
Epistemology Seminar
University of Arizona

Keith Lehrer anchors the pursuit of knowledge and wisdom in our trustworthiness, and in our acceptance and preference of our trustworthiness. Acceptances of beliefs and preferences of desires constitute what Lehrer calls "metamental ascent,” a distinguishing human characteristic. Yet we human beings are also distinguished, less laudibly, by our capacity to deceive ourselves. Mark LeBar1 uses William Talbott’s analysis of self-deception2 to raise two challenges to an epistemology of metamental ascent.3 Talbott claims that, in certain circumstances, an agent will inevitably deceive herself. As LeBar indicates, these circumstances might obtain when the agent seeks to evaluate her own trustworthiness: She would induce herself to accept that she is trustworthy, though she is not. Thus, the spectre arises of an indefeasible competitor to the agent's acceptance of her own trustworthiness, the competitor of self-deception. Lehrer’s entire architecture seems fragile. As LeBar says, “Given the import of my trustworthiness and of the keystone loop it sustains, it would appear worthwhile to ensure that self-deceptive interactions of at least some beliefs and desires do not damage the keystone.… ”4 LeBar calls this implication of Talbott’s analysis the “specific” challenge to Lehrer’s epistemology.

LeBar also identifies a “general” challenge. The demons, brains in vats, and so on, that skeptics have classically invoked as competitors to any acceptance whatsoever, are now joined by the relatively mundane, but ever—present, possibility of our daily self-deceptions. “Appearances to the contrary, Susan really loves me, just secretly;” “I’ll do income taxes early this year, during Christmas break;” “I’m not bald yet, really.” LeBar believes that Lehrer's theory rather readily addresses the general problem. He proposes that, as metamental ascent proceeds, we increasingly consider “the big picture” in our evaluation of beliefs and desires, including the effects of desires, and the effects of preferring desires, on the evaluation system itself. Acknowledgement of the potential corrosiveness of preferring self-deception will tend to constrain such preferences and encourage preferences of trustworthiness. “Given that I prefer to accept that I am trustworthy, at some point the regress of questioning my preferences and possible self-deception in what I accept about them will reasonably stop.”5

LeBar is right about the adequacy of metamental ascent to mitigate the general challenge, but I also believe that metamental ascent mitigates the specific challenge as well. This latter adequacy can be seen in two steps, by first, noting two important characteristics of Talbott's analysis, and second, developing the relation, in Lehrer's system, between a preference for autonomy and our capacity for domain transcendence. In the first step, I will argue that truth is an object of love, of either desire or preference, on Talbott's own analysis, though he explicitly denies that it is an object of desire; and that the process of self-deception that Talbott posits actually presupposes the truth of Lehrer's notion of metamental ascent. In the second step, I will propose that a preference for autonomy motivates domain transcendence, our capacity to trustworthily accept that we are (sometimes) navigating domains in which we are untrustworthy. Since we can trustworthily achieve such acceptances of our untrustworthiness, the competitor at issue—"You deceive yourself regarding your trustworthiness."—is less reasonable than its alternative—"I correctly apprehend my trustworthiness."

The Love of Truth and Metamental Ascent

Talbott seeks to develop an understanding of self-deception that does not presuppose what he calls a "divisionist" conception of the self. A divisionist conception of the self involves a partition of the self that goes beyond the divisions that are customarily posited in accounting for mental behavior that is not self-deceptive. Talbott takes no exception to, for example, reliance on the distinction between our conscious and our subconscious mental lives; but he does resist positing two selves, one of whom, in cases of self-deception, could be construed as "lying" to the other. The goal, thus, is an anti-divisionist metaphysic that can account for self-deception.

The goal seems commendable; indeed, much of the supporting analysis illuminates the nature, and underscores the insidiousness, of self-deception. The anti-divisionist metaphysic posits a single, essentially unitary, self, which is counterposed to no other self, and who, thus, neither lies nor is lied to, neither deceives nor resists deception. Defensible though this metaphysic may be, Talbott formulates the position problematically.

There is no simple, truth-loving victim of self-deception. Because one's cognitive processes are the product of evolutionary selection, they are relatively reliable (at least oriented toward the truth) in their basic operation. Their reliability does not depend on there being a simple lover of truth to guide their operation. They just are reliable. Thus, on my view, we can explain the behavior of the fox [who, the story goes, has deceived himself] solely by reference to the fox's desire to believe that p, regardless of whether it is true. That desire is opposed not by another desire (e.g. the desire to believe the truth about the grapes), but by the ordinarily reliable function of the fox's cognitive processes.6

A problem with the passage will become apparent, a problem that concerns the disavowal of a person, a "victim," who is truth-loving. But, first, Talbott's analysis requires further explication. In support of the metaphysic, Talbott postulates two epistemic principles that LeBar has helpfully summarized as follows.

Epistemic Coherence Condition (ECC): No agent can maintain a belief in the face of the knowledge that that belief has arisen through processes that are unreliable. If its epistemically discreditable origins come to light, its acceptance will end.7

Corollary: I am so constituted cognitively that, given my preference to accept A regardless of its truth, and given the ECC, the nature of my objective in accepting A (the basis of my acceptance of A) will not be apparent to me.8

Now, suppose in instance of self-deception.

SDA: I accept A on some basis other than that it is true, because I prefer to do so.

Insightfully, Talbott proposes the following epistemic scenario.



INTENTION TO BIAS ONE'S COGNITIVE PROCESSES IN FAVOR OF p, which, if successful, produces or sustains:



DESIRE NOT TO BELIEVE THAT q, REGARDLESS OF WHETHER q IS TRUE (where q=that one's believing that p is due to a desire to believe that p regardless of whether p is true) is a product of the level-1 desire to believe that p regardless of whether p is true. It gives rise to:

INTENTION TO BIAS ONE'S COGNITIVE PROCESSES IN FAVOR OF - -q, which, if successful, produces or sustains:



DESIRE NOT TO BELIEVE THAT r, REGARDLESS OF WHETHER r IS TRUE (where r=that one's not believing that q is due to a desire not to believe that q regardless of whether q is true) is a product of the level-2 desire not to believe that q regardless of whether q is true. It gives rise to:

INTENTION TO BIAS ONE'S COGNITIVE PROCESSES IN FAVOR OF -r, which, if successful, produces or sustains:


Thus, a self-deceiver drives herself up a desiderative and cognitive ladder that veils the object of cognition. If the object is her cognitive capabilities, as Lebar notes, she conceals her own untrustworthiness. There are two problems, however, with this application of Talbott. The first concerns a problem within Talbott's analysis itself. Talbott has disavowed that there is a "simple, truth-loving victim of self-deception," that a self-deceptive desire is opposed by a truth-seeking desire." Granted, there is no simple truth-loving victim of self-deception, since the deceived is also the deceiver, and thus not a lover of truth simpliciter. But there is a lover of truth, albeit a lover dupliciter. There is a desire for truth that opposes the desire to believe p regardless of whether p is true. Talbott, without argument, has assumed that truth-seeking derives from evolutionary processes, and thus does not derive from personal agency, including the agency that is motivated by personal passion. At the same time, inconsistently, he is willing to posit a passion for believing that p regardless of whether p is true, even though such passion, doubtless, derives as thoroughly from evolutionary processes as does our tendency to seek the truth. In his effort to avoid divisionism, he attributes passion to our self-deceptions, but not to, for example, our self-apprehensions.

This first problem, within Talbott's analysis itself, relates to a second problem, a problem associated with using Talbott to call into question our trustworthiness, which is crucial to the metamind. One might be tempted to construe the possibility of self-deception as an alternative to metamental ascent. That we often deceive ourselves, so the thinking might go, entails that we cannot count on ourselves. It is true, there are circumstances in which we cannot count on ourselves. But even at worst, this possibility gives us no reason for renouncing the metamind. Quite the contrary, Talbott's analysis essentially presupposes the metamind.

The presupposition appears at each of Talbott's self-deceptive levels.

At Level-1, I must bias my cognitive processes or, consistent with an agent that desires the truth, I will believe p only if it is true. At Level-2, I must believe that -q, or, again consistent with an agent that desires and pursues the truth, I will relinquish my belief that p. The belief that -q constitutes an effort to elude the monitions of the metamind. At the next level, the belief that -r is enlisted to counter the metamind's persistence in accepting only what is worthy of accepting. Were there no metamind, there would be no need for Talbott's self-deceptive agent to generate ever higher levels of corroborating beliefs.

Thus, Talbott's analysis of our capacity for self-deception presupposes Lehrer's conception of metamental ascent; moreover, conversely,the metamind is needed precisely because we can deceive ourselves. Thus, I think, the two analyses presuppose one another. This logical symmetry suggests a further thought regarding the aforementioned problem within Talbott's analysis, the asymmetrical notion that we can desire to deceive ourselves but such desires are not counterposed to truth-seeking desires. Not only has Talbott not argued for this asymmetry; that our capacity for self-deception presupposes the metamind suggests, if anything, that our passional natures are truth-seeking. If there is a metamind—an examination of Talbott himself suggests that there is—why would it be less passional than the self-deceptive tendencies that it serves to monitor and moderate? Lacking passion, the metamind would seem impotent. Clearly it is not… 

Preference for Autonomy and Domain Transcendence

Yet potency is a matter of degree. To what extent can the metamind's system of acceptances and preferences counter our tendency to deceive ourselves? How is the metamind's potency enhanced? The former question, which involves problems of measurement, I do not want to address here; I do want to comment on the enhancement of the metamind's potency. An important dimension of overcoming self-deception—perhaps the essence of overcoming self-deception—might be called domain transcendence, the capacity to trustworthily determine domains wherein one is, precisely, untrustworthy. Domain transcendence is nurtured through a preference for autonomy. 

Consider a common case of trust versus distrust, say, the trust or distrust of a physician. As the one who would trust another, I must trust myself sufficiently to trust another whose knowledge extends beyond my own. I choose a physician on my own authority. She offers advice that I could not have offered myself, indeed, possibly, advice that I hardly understand; whereupon I must decide whether to accede to her advice. On the basis of my own authority—or self trust—I either accept and prefer her advice, acting on it, or seek a second opinion. My self-trust is fraught with paradox: I trust myself to trust someone whose domain is one in which I distrust myself. If, again paradoxically, I seek a second opinion, because I come to distrust the advice of one whose advice I trusted myself to seek, the paradoxical process repeats itself.

Epistemologically, the process is common, even though the degree of human drama varies from instance to instance. The process comprises an unexpungable aspect of human life itself. Due to its everyday character, it may seem unremarkable. It is not unremarkable. It involves acceptances and preferences whereby we initiate further acceptances and preferences that so exceed our ken that, far from being able to offer the advice to ourselves trustworthily, we may not understand the details of the advice, or even know of the details. We may not even wish to know of the details. To wit, the details of our impending urological surgery. But we trust, and proceed to the operating room.

When we trust authorities, the possibilities for deception abound. The authority may intend to mislead us; or she may not intend to mislead us but do so inadvertently, by misspeaking or waxing technical; or she may mislead by being wrong. Nevertheless, we trust ourselves to trust authorities in domains in which we cannot trust ourselves; and, in the very domains wherein we trust ourselves to accept that we are untrustworthy, we trust ourselves to distrust authorities whom we had trusted, trusting ourselves to select other trustworthy authorities. Strikingly, our acceptances often, though not always, facilitate acquisition of truth, our preferences the acquisition of merit. We tend to survive, often prevail. This pleasant fact might be called cognitive domain transcendence in our interpersonal lives.

Our intrapersonal lives also exhibit cognitive domain transcendence. Sometimes I trust myself to distrust myself. Regarding certain matters, say, of love, or of finance, I may have found myself, historically, to be untrustworthy, inclined to deceive myself. In such contexts, I seek a second opinion. Now, it is well and good, a skeptic my say, that I seek a second opinion, or more precisely, should seek a second opinion; but what assures that I will? Well, in fact, there are no guarantees. Indeed, an epistemology that suggested that there were guarantees would virtually refute itself via absurdity. But the likelihood of appropriately distrusting ourselves is enhanced, I think, both by resolving our conflicts through our preferences, and by our preferring that our preferences be autonomous. 

This proposal essentially combines two proposals that Lehrer has made in his discussion of autonomy. He notes two kinds of ways in which we may act heteronomously, internal ways and external ways.9 Internally, my addictions or habituations may impel my behavior so that I want to do, and do, what I prefer that I not do; externally, my desires may be manipulated by another—indeed, I think Lehrer would acknowledge that my preferences may be manipulated by another. The solution, Lehrer says, to the internal problem is "… to require that I do something, not because I want to do it, but because I prefer to satisfy the desire to do it."10 Self-deception would seem to be an internal problem; thus, if Lehrer is right, we should be able to overcome our desires to deceive ourselves by requiring that our acceptances accord with our preferences. Specifically, I may desire both that (A) I accept that Susan loves me regardless of whether she loves me, and (B) I accept that Susan loves me if, but only if, she loves me. Through higher order evaluation I would resolve the conflict. Which, in fact, I would prefer would depend on specific aspects of the situation, though, in most cases, my interests would be better served by preferring (B). Thus, Lehrer's approach to addressing the internal problem carries us a considerable way in enhancing our capacity to counter our tendency to deceive ourselves.

An important problem remains, however. What if we deceive ourselves regarding these very preferences? Suppose I say that I prefer (A) above, not because of peculiarities in circumstances—e.g., I know I will die tomorrow, so it is, in fact, preferable to fall asleep accepting that Susan loves me even if she does not—but rather because my desire that she loves me dominates my attempts at higher order evaluation? In such a case, I may deceive myself regarding my preference, and thereby deceive myself regarding my acceptance. 

The antidote to such a possibility is Lehrer's proposal for addressing the externalproblem, namely, a preference for autonomy. Self-deception might seem to be a thoroughly internal phenomenon; nevertheless, it involves a problem of autonomy. That it involves such a problem can be seen by noting the role that desires serve in Talbott's scenario of self-deception. The desires that Talbott invokes are effectively subconscious or unconscious: At each step, the agent induces herself to accept that her beliefs result from their apparent truth rather than from the desire to accept them irrespective of their truth. The role that the desires serve is thus unconscious or subconscious. When desires unconsciously or subconsciously dominate my preferences, as in this scenario, there is an important sense in which I am not the author of my preferences. The point is not that someone else is the author—someone other than Jones does not author his preferences just because he does not. Yet, neither is it the case that Jones authors them. We might be inclined to gloss the situation by saying that Jones's unconscious or subconscious authors his preferences. But this formulation falters as well, since the notion of "unconscious author" lacks perspicuity and may be incoherent. Further analysis might enable a positive formulation; at this point, I will only express the idea negatively: Jones does not author his preferences when they result from desires unconsciously or subconsciously.

Granted that a literal analysis would be preferable, figurative language perhaps suffices. Just as I guard against literal external manipulation by preferring that I have authored my preferences, so, by preferring that I have authored my preferences, by preferring that I am autonomous, I guard against the figurative external manipulation of self-deceived preferences. Staunch anti-divisionists such as Talbott might take umbrage at this formulation, in which the self as deceiver is construed as external, albeit figuratively, to the self as deceived; but even Talbott acknowledges a division between our conscious and our unconscious lives, and since the notion of "unconscious author" seems less than coherent, we can say that we have not authored our preferences when they arise through unconscious processes. By contrast, when I prefer my autonomy, I prefer to author my preferences, and am inclined to not deceive myself.

LeBar has helpfully noted that self-deceptive processes can curtail our trustworthiness. If our tendencies toward self-deception altogether elude redress, trustworthiness would be discredited as a keystone of coherence. Examination of Talbott, however, reveals that the very epistemic ladder of self-deception, which he so articulately enunciates, presupposes a metamind that impels the self-deceiver to progressively mount the ladder, in an effort to avoid the metamind's continuing scrutiny. Moreover, the scrutiny is enhanced by a preference for autonomy. Thus, if someone accepts that she prefers autonomy, and if she accepts that she is trustworthy, it is more reasonable for her to accept that she is trustworthy than for her to accept that her trustworthiness is the figment of self-deception.


  1. (Back to text)“Self-Deception and the Keystone Loop,” seminar paper, Theory of Knowledge, University of Arizona, fall, 1994.

  2. (Back to text)Forthcoming in Philosophy and Phenomenological Research.

  3. (Back to text)The Keystone Mind (forthcoming).

  4. (Back to text)LeBar, p. 3.

  5. (Back to text)LeBar, p. 6.

  6. (Back to text)Talbott, p. 7.

  7. (Back to text)Summarized by LeBar, p. 4.

  8. (Back to text)LeBar, p. 9.

  9. (Back to text

  10. (Back to text)Ibid., p. 25.

EssaysBrian Flatgard