2. A misconception about non-intentionalism

The answer to the explanatory question on offer here is a non-intentionalist one, but before we develop this there is a widespread assumption about non-intentionalism we must be disabused of. Again, the intentionalist answer asserts that self-deceivers end up believing that p by intentionally making themselves have the belief, a belief they know is false/unwarranted.Footnote ⁵ So non-intentionalism is simply the negation of this: it is just the denial that self-deceivers intentionally make themselves have a belief that they know is false/unwarranted.

As we’ll see, this leaves the non-intentionalist with a variety of explanatory options to pursue. However, the impression is often given in the literature that non-intentionalism is committed to the view that intentional action, motivatedFootnote ⁶ by the desire that p, plays no role in the explanation of the self-deceptive belief. That is, it gets assumed that ‘non-intentionlism’ means ‘no intentional action’. For instance, Bermúdez, a critic of non-intentionalism, says that ‘[a]ccording to anti-intentionalist accounts … self-deceiving belief formation can be explained simply in terms of motivational bias, without bringing in appeals to intentional action’ (Reference Bermúdez2000, 309). Similarly Talbot, a fellow critic, interprets ‘anti-intentionalism’ as holding that the desire causes the belief by ‘triggering’ or ‘activating’ ‘non-intentional mechanisms’ (Reference Talbott1995). And Deweese-Boyd writes that ‘non-intentionalist approaches … [render] the process by which [one] is self-deceived subintentional’ (Reference Deweese-Boyd2006/2012).

On this way of distinguishing intentionalism and non-intentionalism, the theoretical divide between these two persuasions appears very wide. Moreover, since the distinction is supposed to exhaust all possible answers to the explanatory question, it polarizes the debate and hampers our ability to conceive of the full range of explanatory options available to us.

However, the supposition that the self-deceptive belief exists due to the person’s actions (including mental actions), actions which are intentional and motivated by the desire that p, can be consistent with non-intentionalism, so long as the intention these actions are done with is not an intention to deceive oneself (this was all the non-intentionalist was concerned to deny!). Again, this point seems to get missed, even by non-intentionalists themselves. Thus, the well-known non-intentionalist, Alfred Mele, calls the view that self-deceptive beliefs are intentionally acquired/retained, the agency view, and the view that they are not intentionally acquired/retained, the anti-agency view (e.g. Reference Mele1998b, 353–354). But the latter view need not imply that the subjects’ actions are not responsible (partly or fully) for the acquisition/retention of the self-deceptive belief, as long as those actions are not done with the intention to acquire/retain that belief. So it is misleading to call this ‘the anti-agency view’.

The point may be expressed in terms of an important distinction that the remarks of Bermúdez and company hint at:

This distinction, which should not be conflated with the intentionalism/non-intentionalism distinction, can be better understood by noting two different explanatory roles that desires (as in states of desiring) can have.

Perhaps the most commonly recognized explanatory role which desires have is in ‘rationalizing’ explanations of actions, especially intentional actions or actions done for reasons. As Davidson (Reference Davidson1963) noted, when desires feature in such explanations, they help make rational sense of the action, making it intelligible why it was done, helping us to see what the attraction or appeal was in doing it from the agent’s perspective. The most familiar case is where an action is made intelligible by the agent’s desire by being believed to be a means to its satisfaction (e.g. Jeremy shakes the apple-tree because he wants an apple, and believes that by doing so he’ll get one). Mele (Reference Mele1998a) calls this ‘instrumental rationalization’. Note that desires can rationalize other psychological states/attitudes, and not just actions (see Ibid).

But desires and other affective attitudes can enter into other sorts of explanations, where they are referred to as having caused some effect which they don’t rationalize, generally a bodily or psychological change of some sort as opposed to an action or other attitude. Examples might include a desire for food causing one to salivate at the sight of it, worry or anxiety causing a rise in blood pressure, or suffering a heart-attack from terror or fright.Footnote ⁷ Here, the explicanda are ‘mere happenings’ which don’t need to be made intelligible to us as do the explicanda of rationalizing explanations. There is no demand to ask, ‘What was he after?’ or ‘What good did he see in doing that?’, for they are things that happened to the subject, rather than things he did.

I wish to avoid getting bogged down in elucidating this distinction more thoroughly, since this interesting yet daunting task would divert us from our primary concerns. But I hope it’s evident that there is a genuine distinction to be noted here. (Incidentally, I suspect that philosophers who speak of desires as ‘motivating’ actions would reserve this term for cases where desires feature in rationalizing explanations. Thus, they would say that Jones’ desire for food motivated his efforts to get some, but not that it motivated his salivating, rather, it merely caused that.) Furthermore, I suggest that we should think of the desire as playing a rationalizing explanatory role in agentive explanations of self-deception, and a non-rationalizing or ‘merely causal’ one in non-agentive explanations.

Returning to the intentionalism/non-intentionalism distinction, we can now say that intentionalism is clearly a form of agentivism. Non-intentionlist theories, however, can be non-agentivist or agentivist. They are agentivist if they explain the belief as resulting from the subject’s intentional actions, actions rationalized by the desire that p, though the associated intention should not be as intentionalists describe it. They will be non-agentivist if they explain it as resulting from ‘sub-intentional mechanisms’ ‘triggered’ by that desire. (The relationships between these distinctions, regarding their combinability, are outlined in box 1).

Box 1

Note that a non-agentivist explanation, as I am defining it here, need not imply that intentional actions are in no way involved in the causation of self-deceptive beliefs. Suppose S desires that p, but also needs to come to an accurate judgment on whether p and is determined to do so. He searches for evidence both for and against p, motivated by the desire to ascertain the truth. However, suppose that S’s desire that p, unbeknownst to him, primes memories evidentially conducive to believing that p, making them more accessible, a ‘merely causal’ effect of the desire.Footnote ⁸ These memories then come to mind more easily, and he ends up believing that p despite his best intentions, where his actions, say, would have led to him correctly believing that not-p were it not for the operation of this biasing mechanism. Here, the explanation is non-agentivist, even though his evidence search was an intentional action (done with the intention of determining whether p), and was crucial to him forming the belief. This is because his actions or intentions were not to blame for the biased belief, rather, a sub-personal priming mechanism was to blame. His actions would have led to him having the correct belief were it not for the mechanism.

The question of whether self-deceptive beliefs are generally due to the person’s actions or to the operation of sub-personal/sub-intentional mechanisms is important, with major consequences for the issue of to what extent people are responsible for their own self-deception. Many non-intentionalist theories emphasize the mechanistic (e.g. Hales Reference Hales1994; Lazar Reference Lazar1999; Smith Reference Smith2014). This, however, is an over-reaction to intentionalism’s objectionable features, and I here wish to convey the attractions of agentive non-intentionalism, a frequently overlooked theoretical possibility. First, this approach explicitly endorses the common belief that there are self-deceptive actions (Hellman speaks of our ‘intuitions that self-deception is a state determined by our behavior’ (Reference Hellman1983, 120). Also see Scott-Kakures (Reference Scott-Kakures2002, 591) and Michel and Newen (Reference Michel and Newen2010, 742)). The grammar we use when talking about self-deception suggests this (Szabados Reference Szabados1974b, 62), as when we use the active progressive ‘She is deceiving herself’, which communicates that the subject is actively involved in her self-deception, as opposed to the stative ‘She is self-deceived’, which does not directly suggest this. But it endorses this point while avoiding intentionalism, with its threat of paradox.

Furthermore and relatedly, it accommodates the common intuition (e.g. Fernández Reference Fernández2013; Szabados Reference Szabados1974a) that a person’s self-deception is something she can be held responsible for. The idea that we end up self-deceived due to our own actions seems most congenial to this assumption (in that we can often be held responsible for the consequences of our actions). Note, however, that the current theory will be agentivist in a qualified sense, since it takes the self-deceptive belief to result from the subject’s actions and omissions. But this will not undermine its accommodating our intuitions about responsibility, since it’s undisputed that people can be blameworthy for their omissions. Alternatively, if self-deception is more the result of the automatic, ‘merely causal’ influence of desire on cognition, we would perhaps be better thought of as hapless victims of the condition, rather than perpetrators.

However, perhaps the main reason for favoring an agentivist non-intentionalism is that empirical studies of self-deception (i.e. cases displaying features (1)–(3)) suggest that the ‘mediating processes’, as psychologists say, between the desire and belief involve intentional actions.Footnote ⁹ Next, a relevant representative study will be presented and interpreted in this fashion.

3. Mediating processes between the desire and belief

In the studies I have in mind, typically a large group of subjects would be presented with evidence suggesting that a proposition is true. Some of them would very much not want it to be true (call them ‘stakeholders’), whereas the others wouldn’t care either way (‘non-stakeholders’). In one such study by Liberman and Chaiken (Reference Liberman and Chaiken1992), a replication of one by Kunda (Reference Kunda1987), subjects were told that they were participating in a study on ‘nonscientists’ understanding of scientific and technical information’ (Reference Liberman and Chaiken1992, 672). Then they were given an article to read purporting that moderate to heavy caffeine consumption greatly increases the risk of fibrocystic disease, a serious breast ailment associated with cancer. The subject pool consisted of women sorted into two categories: ‘low-relevance’ subjects, who drank no cups of coffee per day (making them non-stakeholders), and ‘high relevance’ subjects, who drank two to seven cups per day (stakeholders). After reading the material, they filled in questionnaires about it.

From examining these questionnaires, the experimenters found that, as expected, stakeholders tended to be more skeptical of the proposition compared to non-stakeholders. On average their opinions veered in the direction of what they wanted to be true relative to non-stakeholders’ views.

Measures were included on the questionnaires to determine the cause of this bias. The article contained four research reports on whether caffeine is linked to this disease, one anti-link report and three pro-link reports. Methodological flaws or weaknesses were included in each report ‘[to] allow vigilant subjects the opportunity to be critical’ (Reference Liberman and Chaiken1992, 672). Subjects were then asked to list any strengths and weaknesses that they found in these reports, and their answers to this could be used to assess how the evidence was being treated.

Generally speaking, there are two ways a self-deceiver could respond defensively to unwelcome evidence: she could make a flight response, attempting to avoid or ignore it, or a fight response, attempting to refute it. These two possibilities were tested for. First, Liberman and Chaiken checked for a defensive inattention hypothesis: that stakeholders ended up with their biased views by avoiding attending to the pro-link sections while attending more to the welcome anti-link sections (a flight response). From answers given to a memory test on the details of the reports however, it was found that this was not what happened, a result consistent with other studies. Instead, they found evidence for what they dubbed biased systematic processing. Stakeholders attended to both kinds of report equally, but were discovered to have rated the anti-link report as superior. Crucially, whereas non-stakeholders found an equal amount of weaknesses in pro- and anti-link reports, stakeholders listed ‘significantly more’ weaknesses in the pro-link reports and found less weaknesses in the anti-link report, compared to non-stakeholders (Reference Liberman and Chaiken1992, 674). As the authors conclude, ‘[c]ompared with low-relevance subjects, high-relevance subjects were less critical of those parts of the message that were reassuring and more critical of those parts that were threatening’ (Reference Liberman and Chaiken1992, 675). They exhibited a fight response to the unwelcome evidence, attempting to refute/discredit it.

This study suggests that it was a combination of the subjects’ actions (being hypercritical of/looking for flaws with threatening evidence) and omissions (failing to be critical of reassuring data) that led to the formation of the biased beliefs. In other words, their beliefs resulted from their acts of searching specifically and one-sidedly for such considerations (flaws) that they would like to find. This is consistent with results found in numerous other studies (e.g. Ditto et al. Reference Ditto, Munro, Apanovitch, Scepansky and Lockhart2003; Ditto and Lopez Reference Ditto and Lopez1992; Frey and Stahlberg Reference Frey and Stahlberg1986; Holton and Pyszczynski Reference Holton and Pyszczynski1989; Lundgren and Prislin Reference Lundgren and Prislin1998; Pyszczynski, Greenberg, and Holt Reference Pyszczynski, Greenberg and Holt1985; Wyer and Frey Reference Wyer and Frey1983). For instance, Frey (Reference Frey1981) found that subjects led to believe that they had performed poorly on an intelligence test showed a preference for reading articles critical of intelligence tests over articles supportive of them. (Note that an evidence search may be physical, as with Frey’s subjects looking through these articles, or it may be mental, involving just trying to think of arguments or reasons in support of a proposition.)

Now I don’t deny that other non-intentionalists, such as Mele and Scott-Kakures, would agree that this goes on in self-deception, and they may even have mentioned similar things in their work. However, they have not supposed, or explored the idea, that this might be evidence for a distinctive kind of intention present in self-deception. Identifying this intention promises us a deeper explanation of the phenomenon (as we’ll see better in Section 5).

The argument that the culpable acts evident in these studies are intentional is a simple and, I hope, a compelling one. First, it is a general truth that acts of searching are goal-directed and intentional. They are done with the intention, or with the hope, of finding whatever one is searching for. Thus, if you are searching through your pockets for your key, then you intend to find your key, or if a detective is searching for a missing person, he acts with the intention (or hopeFootnote ¹⁰) of finding that person. Now the stakeholders in the study seemed to be engaged in a search. But what were they searching for exactly? Apparently, for flaws in the threatening evidenceFootnote ¹¹ (they found many flaws in the pro-link reports but detected few in the anti-link reports, though they were there to be found). Are we not compelled, then, to attribute to them something like the intention to find any flaws in the threatening evidence? Note, however, that although the action component of our explanation is intentional, the omission component need not be, a point that will soon prove useful for rebutting accusations of paradox.

In Section 4, this idea will be developed further, and we will see how the desire that p can give rise to such intentions. The rationalizing role that this desire has in explaining these intentions will also be clarified, which will help make the phenomenon more intelligible to us, and we will address the question of whether there is any danger of paradox with this account.

4. The associated intention

The above study and others like it suggest that self-deceptive beliefs result, at least in part, from one-sidedly seeking evidence/considerations that would support one’s favored view, while failing to submit such reassuring considerations to critical scrutiny. Note, however, that an intentionalist could accept all this and subsume these findings into her own theory. Accordingly, she could say that self-deceivers search one-sidedly for welcome, p-supporting evidence with the intention of making themselves believe that p, a belief they initially know is false/unwarranted. They want to believe that p (to avoid the anxiety of believing/suspecting that not-p), think that by searching for evidence/considerations in this biased way they may achieve that, and then act like so to achieve it.

This interpretation of the self-deceiver’s behavior would be thought by many to invite paradox. Some would think that it implies that the subjects would at one point concurrently believe that p and that not-p, though rarely does one see an argument for why the beliefs couldn’t be held consecutively instead.Footnote ¹² The more serious threat of paradox comes from the idea that they intentionally and knowingly search for evidence in a biased way, for how could they be persuaded by a body of considerations that they know is biased? Now perhaps the intentionalist has a reply to this, but I won’t pursue this debate. Instead, an indirect argument against the intentionalist will be provided, by showing that an alternative understanding of the self-deceiver’s behavior is available which is preferable because it is simpler, psychologically more plausible, and paradox-free. The rationalizing role of the desire that p in this account will also be outlined.

On this alternative non-intentionalist theory being recommend, all we need to say to account for the subject’s biased behavior, call him Jones, is the following. After the unwelcome evidence indicating that not-p is encountered, Jones finds that a previous assumption of his is threatened with being false (though he does not form the belief that not-p on its basis). Because of his strong desire that p, this prospect is a source of distress for him. Because of this, Jones has a heightened interest in any evidence/reasons there might be that would invalidate that unwelcome evidence and remove the threat, for the reason that it would put his mind at ease to have p confirmed. He consequently becomes anxious to find such considerations, which, again, if found would allow him to rest assured that p. Anxiously desiring to find reasons to discount that evidence, he acts with the intention of satisfying this desire, and searches for such considerations. Thus, he acts merely with the intention of finding any weaknesses with the threatening evidence. Or perhaps he acts with the broader intention of finding any considerations supportive of p (which would include flaws with the evidence against p, but could also include considerations independently supportive of p). Thus, he is searching for considerations he would welcome or like to find, though his searching need not be intentional under that description.Footnote ¹³ Moreover, in his eagerness to find p-supporting considerations, he neglects to submit those considerations to critical scrutiny.

On this account, the culpable behavior (one-sidedly searching for welcome considerations) is explained by an intention (to find any problems with the threatening evidence, or to find any p-supporting considerations), and that intention is explained and rationalized by the desire that p along the above lines. Again, it should stir no controversy to say that searching for flaws with the threatening evidence, or for p-supporting considerations, implies an intention to find such things, just as searching for, say, a missing sock implies having an aim or intention to find that sock. Thus, intentional action is a key component of our account.

Now it may be observed that with the intentionalist interpretation, Jones acts with this intention also. On that view, Jones intends to find p-supporting considerations too, but the difference is that he is supposedly doing this in the attempt to satisfy his further intention to believe that p. What the current view is saying is that the intention to find any p-supporting evidence need not be seen as a sub-intention in this larger intentional project for Jones’ behavior to be intelligible to us. For just in light of the fact that he desires so much that p we can understand why he so eagerly seeks such considerations, simply because his finding them would assure him that what he wants to be true is true, thus putting his mind at ease. This account is therefore simpler than the intentionalist’s.

Importantly, this allows us to see how the desire that p rationalizes the self-deceptive actions, how it ‘makes sense’ of them, showing that the desire is not playing a ‘merely causal’ or mechanistic explanatory role. It is fully intelligible to us why Jones doggedly seeks p-supporting evidence, since given that he desires that p, it would please him to find it; it would be to his relief (a case of non-instrumental rationalization; see Mele (Reference Mele1998a)). Furthermore, ‘cold biasing mechanisms’ do not come into this picture. The biased evidence search is not a case of the confirmation bias being triggered by the desire that p. Rather, these subjects are searching for p-supporting evidence because they want to find such evidence, to assure themselves that p is true. The activity is both initiated and sustained by this desire (thus satisfying a desideratum of Scott-Kakures (Reference Scott-Kakures2000)).

There should be no mistaking how different this account of the self-deceiver’s intention is from intentionalism’s. Bermúdez (Reference Bermúdez2000, 310) has described various options for how intentionalists can conceive of the self-deceiver’s intention, but they all involve the self-deceiver intending to bring it about that he acquires a belief. On this view, the self-deceiver’s concern is directed inward (figuratively speaking): he is concerned with his own doxastic state. We flatly deny that the self-deceiver has any such intention, and shift toward seeing the self-deceiver as having an outwardly directed concern. By this we mean that he is concerned with the truth-value of p, not with his own doxastic state. For he is, to reiterate, anxious to put his mind at ease that it’s actually true that p, and so acts in the hope, and with the intention, of finding any considerations that would confirm this, not in the hope of making himself have a belief. This is sufficient to explain the biased behavior visible in the empirical studies, and because he does not have an intention to make himself have a belief, our account avoids any danger of paradox associated with that idea.

The idea that the self-deceiver has an outwardly directed concern makes this account, in my view, more psychologically plausible than the intentionalist’s. For us, the self-deceiver’s primary motivation is to find evidence that would confirm p. For the intentionalist, it is to avoid negative affect. But the issue of whether p is clearly the more urgent and significant one. The issue of whether one is likely to get fibrocystic disease, or of whether one’s partner is just after one’s money, is weightier than the issue of what negative affect one may feel if one believes that these things are true. It is more plausible that our anxieties about the former are what generate self-deceptive biases. We will return to this matter in the final section.

It remains to be shown that this account of the self-deceiver’s intention is not blighted with paradox. Intentionalism invites paradox because the self-deceiver is conceived to intentionally act in a deceptive, epistemically suspect manner. Crucially however, on our account, the intentional actions of the self-deceiver are not epistemically suspect. Here, the point that the bias consists in the self-deceiver’s actions and omissions becomes important. Recall that the biased behavior of the self-deceiver consists in the following:

1. (1) Seeking welcome considerations.

2. (2) Not seeking unwelcome considerations (which includes failing to critically scrutinizing any welcome considerations which are found).

This distinguishes self-deceivers from rational, judicious inquirers, who look for the pros and cons, and not just for pros (or cons) only. Now (1) is an intentional action, since acts of searching are generally intentional. So we may suppose that the subject knows what he’s doing here, that is, he knows that he’s looking for problems with the threatening evidence, or considerations which may favor p.Footnote ¹⁴ But there is nothing intrinsically suspect about looking for problems with certain evidence, or looking for consideration which might support a certain proposition. Such behavior does not constitute being biased. The bias consists in doing that, while omitting to critically assess the welcome considerations that one finds or thinks of, and omitting to put an equivalent effort into looking for considerations supportive of not-p. But these omissions need not be intentional. They are instances of neglecting rather than of avoiding or refusing to do something. People naturally put more energy into looking for things they want to find, compared with things they do not want to find. The self-deceiver is just so driven to find p-supporting considerations that he neglects to seek and check for not-p-supporting considerations, or at least neglects to do so with the same energy. Because these omissions are not intentional, there is no pressure to say he must have been aware of his omissions, and so no pressure to say he was aware he was being biased.

Suppose, however, that he did realize that his evaluation of the matter was one-sided. Would that undermine his self-deceptive belief? It might, and it’s plausible to think that reflective self-awareness of this sort is self-deception’s antidote. But it also might not. For he still does not know that he would have found problems with the welcome considerations if he had looked for them. Though he might acknowledge his bias, he may still cling to his self-deceptive belief for lack of having any specific reasons to surrender it.

A worry for this account, however, relates to whether it is faithful to our own experience of when we do such things. When we reflect on occasions when we acted like so, does it not seem that our searching for welcome considerations proceeds sometimes in a more automatic than deliberate manner? Don’t we just immediately seize upon such welcome considerations, without much reflective awareness that we are doing so? It can be admitted that there may be a spectrum of cases here, but the point is one that is true of many sorts of actions. That is, for many actions which are sometimes done in a deliberate or intentional manner, they can at other times be done in a more automatic, spontaneous, or habitual manner (and sometimes in a semi-intentional manner, since these things come in degrees). TV ‘channel surfing’, for instance, may be done in a very deliberate way, or in a more automatic, ‘mindless’ way, though in both cases the action is explained by the same desire to find something good to watch. It shouldn’t be surprising if actions of searching for flaws with unwelcome evidence are no different in this respect.

5. The relation between this and Mele’s processes of self-deception

Before finishing, this account should be related to Mele’s, who has written a lot on how ordinary or ‘garden-variety straight’ self-deception (the phenomenon targeted here) comes about. Through this we will see the explanatory depth of the present account, which primarily consists in its unifying power.

Mele mentions a number of apparently independent processes that may help explain ordinary self-deception. However, it seems that many of these processes can be subsumed into and unified by the present account; they can be seen as manifestations of the one phenomenon. Consider two such processes. He says that self-deceivers may engage in ‘negative misinterpretation’, which is when ‘[o]ur desiring that p may lead us to misinterpret as not counting (or not counting strongly) against p data that we would easily recognize to count (or count strongly) against p in the desire’s absence’ (Reference Mele2001, 26). The example is given of Don who had his paper rejected from a journal, and who forms the belief, unwarrantedly, that it was wrongly rejected because the reviewers misunderstood a certain crucial but complex point. Or they may engage in ‘positive misinterpretation’, which is when ‘[o]ur desiring that p may lead us to interpret as supporting p data that we would easily recognize to count against p in the desire’s absence’ (Reference Mele2001, 26). He illustrates with the case of Sid who interprets the rebuffs of a woman he’s infatuated with as a sign of her playing hard to get.

It seems, however, that both Don and Sid’s behavior can easily be understood as being instances of striving to find p-confirming/supporting considerations (so that they might rest assured that p is true). Don looks for and finds a reason to think that the reviewer got it wrong (as he hopes), a reason he doesn’t scrutinize, while Sid looks for and finds a reason to think that some ostensibly negative evidence isn’t negative at all.

Mele also mentions ‘selective focusing/attending’ which is when ‘[o]ur desiring that p may lead us both to fail to focus attention on evidence that counts against p and to focus instead on evidence suggestive of p’ (Reference Mele2001, 26). This is also easily explicable by us. The reason why the self-deceiver fails to notice evidence for not-p is that he’s not looking for it. He is looking for evidence for p, however, which is why he notices/focuses on that. How Mele’s ‘selective evidence-gathering’ (Reference Mele2001, 27) fits into our account is even more obvious. If one is looking for evidence for p but not for not-p, one will end up selectively gathering evidence for p. Furthermore, Mele suggests that data consistent with a desired hypothesis can be rendered more ‘vivid’ or ‘salient’ by that desire (Reference Mele2001, 29). But again, this ‘vividness’ can be understood as a consequence of the fact that the subject is searching for that data (think of searching through a cluttered drawer for something. The item you are looking for jumps into your attention and you barely notice the other items).

So many of Mele’s self-deceptive processes can be explained in terms of, or understood as manifestations of, the attempt to find any welcome evidence, and moreover, by providing this common explanation, we significantly improve on his account by bringing unity to this otherwise disparate-looking set of factors. We are not denying Mele’s explanations but are deepening our understanding of them, by explaining those explanations. Note also that in doing this, we are assimilating them into an agentive account of ordinary self-deception. We understand these factors not as mechanisms or the effects of mechanisms, but as actions or the effects of actions. Mele, on the other hand, promoted an incorrect understanding of these processes by calling his view an ‘anti-agency’ view (Reference Mele2001, 25, 49).Footnote ¹⁵

It is doubtful, however, that all the explanatory processes that Mele mentions can be subsumed into this account. Mele’s ‘FTL model’ in particular (an acronym referring to the work of psychologists J. Friedrich, Y. Trope, and A. Liberman), which he offers as a possible, partial explanation of garden-variety self-deception, is one we should perhaps take exception to. As this view has become quite popular with non-intentionalists, championed especially by Scott-Kakures (Reference Scott-Kakures2002, Reference Scott-Kakures2000), it deserves a sustained discussion. Next we will see that the FTL phenomenon is distinct from the psychological phenomenon we have been looking at here, and reasons will be given for being skeptical about emphasizing it in the explanation of self-deception.

First, a brief exposition (see Mele (Reference Mele2001) for more detail). The FTL model uses the notion of a ‘confidence threshold’, which refers to the quantity and quality of evidence required before one accepts a proposition as true. It proposes that confidence thresholds vary depending on what the expected costs are of falsely believing the relevant proposition, costs which the subject desires to avoid. The less costs associated with falsely believing that p, the lower one’s confidence threshold regarding it (i.e. it will take evidence of a relatively lesser quality/quantity to make one accept it), and the more costs associated with falsely believing that p, the higher one’s confidence threshold (i.e. it will take evidence of a relatively higher quality/quantity before one accepts it).

Regarding some hypothesis p that a subject attempts to evaluate, the expected costs associated with wrongly settling the question in favor of p or not-p may be ‘asymmetric’ in magnitude. The subject will then be most concerned with avoiding the costlier error, and thus will have different thresholds for coming to believe p compared to not-p. If believing that p falsely would be most costly, then his threshold for accepting that p will be higher, and it will take relatively strong evidence to convince him of that. Consequently, this subject may be biased toward concluding that not-p. He will be disposed to settle on that conclusion more easily. Furthermore, he will be disposed to scrutinize carefully evidence suggesting that p for errors, since he is averse to mistakenly concluding that p.

To illustrate, consider Frank, a lead engineer overseeing the construction of an airplane. Frank has to evaluate whether the plane is flight-worthy. He knows that there are high costs associated with falsely concluding that it’s flight-worthy (death, litigation, and unemployment). He knows that the costs of falsely believing that it’s not flight-worthy are much less significant (resources needlessly wasted). He desires most to avoid the costlier error. So his confidence threshold for accepting that the plane is flight-worthy is higher. Thus, he is disposed to conclude that it’s not flight-worthy more easily, and he will scrutinize carefully any evidence (e.g. reports) suggesting that it is flight-worthy.

The first point to make about such cases is that it is not clear that the thresholds in question here are belief thresholds as opposed to acceptance thresholds in the distinct sense explained by Bratman (Reference Bratman1992) and others. For acceptance is supposed to be influenced by precisely the sorts of practical considerations described in the FTL model. And if they are acceptance thresholds then this might undermine the relevance of the FTL phenomenon to self-deception insofar as self-deception involves belief. So just as, say, if one were driving on a remote and dangerous road one might ‘accept’, ‘presume’ or ‘work on the assumption’ that one is going to meet another car on the blind turns while possibly believing that it’s quite unlikely to happen, the engineer might just be accepting that the airplane is not yet flight-worthy without that necessarily meaning that he believes it’s not yet flight-worthy.

But let us grant, for the sake of argument, that the thresholds are belief thresholds. This example then shows that ‘FTL motivation’ (motivation to avoid doxastic error costs) does not in itself lead to self-deception. For even if Frank ended up with a belief that the plane is not yet flight-worthy when it is, though he might be guilty of over-cautiousness, he would hardly be guilty of self-deception. Certainly this would not be a case of ‘garden-variety straight’ self-deception since this belief may not be one which he desires to be true (it could make little difference to him when the plane becomes flight-worthy). To see how the FTL phenomenon is relevant to self-deception, we must look at how desires relating to the truth of the proposition can give rise to doxastic error costs.

Mele says that where someone has a strong desire regarding the truth of a proposition, this will affect what costs there are for him in falsely rejecting/accepting it. Most importantly, where S desires that p, if S falsely believes that not-p he will incur negative affect (Reference Mele2001, 35–36), but not if he falsely believes that p. S’s desire to avoid this cost will then give him a relatively low threshold for believing that p, and will make him vigilant with regard to weaknesses in evidence supporting not-p.

Note, however, that the thing which S supposedly wants to avoid here, negative affect, would be incurred whether S believes that not-p falsely or truly. Experiencing this affect is not dependent on the belief being a mistaken belief. Mele acknowledges this point (Reference Mele2001, 35), but it implies that this does not exemplify FTL motivation: it is not a matter of being motivated to avoid the costs of making some sort of error. When we realize this, we see that this is just an old idea dressed up in new garb: the idea that in self-deception, the subject is motivated to avoid the anxiety associated with believing something, which biases his reasoning.

This idea that the self-deceiver’s concern is an ‘inwardly focused’ one of avoiding anxiety is one this paper has been trying to move us away from; it conveys an implausible picture of the self-deceiver’s primary worry. For when one, say, encounters evidence suggesting that one may be seriously ill, one does not, first of all, become worried about the anxiety one will suffer in believing this. One’s worry primarily concerns being seriously ill, with all the implications that would have for one’s future (or lack of one). Surely any relatively trifling worries about what negative affect one might suffer in believing this (falsely or truly) will be secondary to this, if they arise at all (Bach Reference Bach1981, 353; Szabados Reference Szabados1974c, 470). And it should come as no surprise if this outwardly directed concern has the power to distort one’s thinking, independently of any inwardly directed concerns about avoiding emotional distress.

It is time that we questioned the dominant yet implausibly solipsistic idea that the self-deceiver’s primary concern is with avoiding anxiety rather than with the truth/falsity of the proposition he’s self-deceived about. For we know only too well that our having a major stake in some proposition being true leads naturally (when the proposition is thrown into doubt) to an intense desire for there to be evidence that it is true, which leads us to search doggedly for such welcome evidence, so that we might be relieved that it is true, and this motivational complex by itself can generate a biased belief. It is this motivational complex that is the main engine of ordinary self-deception, and a person so concerned cannot be rightly described as being concerned about negative affect. She is concerned about whether something is true or not.Footnote ¹⁶

The cost of negative affect is what Mele emphasizes the most, but other sorts of costs have been claimed to be generally associated with falsely rejecting desired propositions. It has been claimed, for instance, that this may often involve missing important opportunities (see Scott-Kakures Reference Scott-Kakures2002, 594). Consider a subject, S, in Frey’s experiment, who convinces himself that intelligence tests are inaccurate measures of intelligence after having performed poorly on one. If S falsely believed that intelligence tests are accurate, he could take himself to be less intelligent than he is. He might then assume that he is unfit for pursuits (e.g. a certain career) he actually is fit for, and could thus miss opportunities. Granted. But note that there are potential costs, no less serious, with the converse error. If he falsely believed that intelligence tests were inaccurate, he might assume that he is smarter than he is. This could lead him to take up pursuits that he’s unable for, leading to eventual failure, disappointment, and perhaps wasted years. Both overestimating and underestimating your intelligence carries risk. Or consider a subject in Liberman and Chaiken’s experiment, who convinces herself that the studies linking caffeine consumption to fibrocystic disease are problematic. Wouldn’t there be significant costs with falsely accepting that proposition, in that she could be dismissing a genuine danger?

Advocates of the FTL model have thought that it could serve as a general explanation of paradigmatic self-deception because it has seemed to them that we generally regard falsely rejecting desired propositions as more costly than falsely accepting desired propositions. But it’s not evident that this is true. Costs can often be identified on both sides. We would thus need evidence that in cases of self-deception generally, subjects associate high (and higher) costs with falsely rejecting the welcome proposition, if there is to be any hope of the FTL theory delivering the goods. Without that, it is little more than a conjecture. The protagonist motivational states in the current explanation, however (e.g. the desire to find problems with the threatening evidence, or to find p-supporting evidence), are more likely to be generally present in the relevant cases (who doesn’t want to find p-supporting evidence when a proposition, p, that one strongly wants to be true is thrown into doubt?). And furthermore, they are likely to be a more powerful force in the subject’s mind than the likes of desires to avoid anxiety or to not miss opportunities, and so are more likely to be the main culprit responsible for motivating the biased behavior.