The IKEA effect and the production of epistemic goods

Abstract

Behavioral economists have proposed that people are subject to an IKEA effect, whereby they attach greater value to products they make for themselves, like IKEA furniture, than to otherwise indiscernible goods. Recently, cognitive psychologist Tom Stafford has suggested there may be an epistemic analog to this, a kind of epistemic IKEA effect. In this paper, I use Stafford’s suggestion to defend a certain thesis about epistemic value. Specifically, I argue that there is a distinctive epistemic value in being an active producer of epistemic goods, like true belief, as opposed to just a passive recipient of such goods, and that because of this it can be rationally permissible to sacrifice truth in a certain way for the sake of this other value. In particular, it is rationally permissible for an epistemic agent to prefer a belief set that contains fewer overall truths but more truths obtained through the agent’s own intellectual labor, in something like the way that a practical agent might prefer furniture they have made through their own manual labor to inherently superior furniture made by someone else. In making my case, I draw on Ernest Sosa’s discussion of causation and praxical epistemic values, and Jennifer Lackey’s testimony-based criticism of the credit view of knowledge. After defending my thesis about epistemic value, I further clarify it by connecting it to the focus of Stafford’s discussion, conspiracy theorists.

Introduction

Behavioral economists have proposed that people are subject to a so-called IKEA effect, whereby they attach greater value to products they make for themselves than to otherwise indiscernible goods. Recently, cognitive psychologist Tom Stafford (2021) has suggested that there may be an epistemic analog to this, a kind of epistemic IKEA effect (see also Vuculescu et al., 2021, for a related line of thought). In what follows, I use Stafford’s suggestion to shed light on a thesis about epistemic value that I want to defend. There is a distinctive epistemic value in being an active producer of epistemic goods, I argue, and not just a passive recipient of such goods. This value can come into conflict with other epistemic values, including that of truth, and when it does it can be rationally permissible to sacrifice truth for this other, productive value, in that an agent can prefer to have a belief set that contains fewer overall truths to one that contains more, provided that the first set contains more truths obtained through the agent’s own intellectual labor. One way to put it: there is a value in thinking for yourself that is comparable to that of making something for yourself.

I begin in Sect. 1 by reviewing the IKEA effect, focusing on elements that set up the epistemic discussion to come. In Sect. 2, I begin developing my thesis regarding epistemic value by drawing on Ernest Sosa’s discussion of praxical value. In Sect. 3, I refine my account by considering Jennifer Lackey’s discussion of testimony and epistemic credit. In the concluding Sect. 4, I attempt to illuminate my thesis about epistemic value further by considering how it might bear on the focus of Stafford’s discussion, conspiracy theorists.

The IKEA effect

The term “IKEA effect” was coined by behavioral economists Norton et al. (2012) for a phenomenon in which consumers attach greater value to products they create through their own labor than to otherwise indiscernible products made by others. The paradigm example and basis for the name is the unassembled furniture sold by the Swedish multinational conglomerate IKEA. When you buy, say, a bureau or a bookshelf from IKEA, you typically have to do the work of putting the item together when you get home from the store. What Norton and colleagues propose is that this labor leads consumers to value these products more highly than they otherwise would. You like your IKEA bookshelf more, precisely because of the causal role you played in putting it together.

The effect is not restricted to furniture. Norton and colleagues begin their paper by recounting that when instant cake mixes were first introduced in the 1950s, as part of a general trend to make housework less onerous (alongside washing machines, automatic dishwashers, and so on), there was initial consumer resistance because the mixes seemed to make cooking too easy, undermining the value of the labor involved. When manufacturers responded to this dissatisfaction by changing the recipes to require the addition of an egg—a bit of cooking labor that the consumer could contribute—the mixes actually became more popular (Shapiro, 2004). Today, cooking is a paradigm example used to illustrate the IKEA effect, and one of the domains in which it is most studied (Dohle et al., 2014; Radtke et al., 2019). To get your kids to eat their vegetables, let them do the work involved in planning and preparing meals that involve vegetables and then sit back and let the IKEA effect do its work. Or at least so it has been suggested (Radtke et al., 2019).

Something like the IKEA effect—although not the name or a detailed understanding of the psychological mechanisms involved—has long been known to marketers, driving a movement to treat customers as creators of value in their own right rather than just as recipients of value. The effect is generally understood to involve a cognitive bias, but spelling out exactly what the bias consist in takes some care. It is not inherently practically irrational or otherwise a mistake to attach greater value to products you make than to ones you don’t. This is a rationally permissible preference to have, and so it alone can’t be what the supposed bias consists in.

I pause over this point because it matters to my project. I will be arguing that there really is a genuine epistemic value in thinking for yourself, so that epistemic agents can rationally prefer opportunities to do so even if it means sacrificing other things of epistemic value, like truth. When I turn in Sect. 4 to suggest that conspiracy theorists might be understood as pursuing opportunities to think for themselves, my claim will not be that this is where they go wrong—that it is irrational or otherwise a mistake that they dare to think for themselves. In doing so, they are pursuing a genuine form of epistemic value on my account. Instead their error must consist in something else.

Although Norton and colleagues don’t quite put it this way, the bias or mistake in the IKEA effect involves something like overestimating the intrinsic value of the products you make, mistakenly regarding your preference for them as reflecting their inherent superior quality rather than your role in creating them (an extrinsic matter). A primary manifestation of this bias involves wrongly supposing that others will value your products as much as you do. In one experiment, Norton and colleagues found that subjects tasked with making origami frogs and cranes viewed their constructions as being nearly equal in value to origami made by experts, and expected others to be willing to pay accordingly. In fact, others viewed the origami figures as “nearly worthless crumpled paper,” and were willing to pay significantly more for expert work (Norton et al., 2012: 455). In popular presentations, the example is sometimes given of home sellers overestimating what the market will pay for their house based on the renovations they themselves have done, the “sweat equity” they have put into it. “Avoid the IKEA effect” has filtered down from the academic literature and become advice given to sellers (Myrick, 2013).

In the epistemic case, the idea will be that the bias involves overestimating the likelihood of your beliefs to be true (the analog to thinking that your origami frogs and cranes are of high quality) based on the amount of intellectual labor you have personally poured into them. When it comes to conspiracy theorists, this will characteristically involve them taking the belief sets they have constructed to be superior to those produced by experts. We will return to this suggestion at length in Sect.4. At this point, though, I want to turn to my arguments regarding the genuine form of epistemic value involved in the production of epistemic goods.

Sosa on praxical epistemic value

My argument begins by drawing on Ernest Sosa’s (2003) discussion of epistemic value. In that work, Sosa is addressing the primary value problem in epistemology, which has to do with the value that knowledge possesses over and above mere true belief. I mention this to provide context, but for our purposes we can remain neutral on Sosa’s view of knowledge and his proposed solution to the primary value problem. It is possible to endorse much of what Sosa has to say about epistemic value even if we don’t agree with just how he connects it to the primary value problem.1

Sosa (2003: 161–162) begins his discussion with a moral analogy. Suppose that ethical hedonism is true and so pleasure is the sole thing of final (non-instrumental) moral value. Then consider a pair of worlds, W1 and W2, each of which contains events involving two types of pleasure, X and Y. In W1, the instance X1 causes the instance Y1. In W2, the instance X2 is followed in time by but does not cause the instance Y2. The two worlds are otherwise indiscernible. The two worlds contain the same amount of pleasure and so the same amount of final moral value, given our premise of ethical hedonism. Still, Sosa claims, the instance of pleasure X1 is better than the instance of pleasure X2, for although the two instances of pleasure are intrinsically indiscernible—they are pleasurable experience of the exact same type—X1 brings additional pleasure into being by virtue of causing Y1, while in contrast X2 does not bring such additional pleasure into existence, it does not cause Y2. Because of this causal difference, X1 possesses a kind of instrumental value that X2 does not, and so possesses greater overall value than X2, understood as something like the sum of its final and instrumental value.

Sosa uses the scenario to make a parallel point about epistemic value. Suppose that truth is an epistemic analog of pleasure, in that it has final epistemic value. (Falsity meanwhile has final epistemic disvalue.)2 Suppose next that an agent A holds a true belief B. Suppose finally that A as an agent plays a causal role in bringing about B, doing so via an intellectual performance, understood as a causal process or sequence of mental events that involves the manifestation of an intellectual ability or capacity. A arrives at their true belief B in a way that is attributable to them as their own doing; the true belief is not merely something that happens to A. In that case, A earns a kind of intellectual credit for their true belief. To have an example, you might suppose that A carries out a bit of logical reasoning, starting with true premises and arriving at a true conclusion, and that A does so not though luck but thanks to their skilled logical ability.

Sosa proposes that in such a case, there are two interrelated forms of epistemic value that obtain in addition to the final value of B’s truth, two forms of what he calls praxical epistemic value. We can formulate his proposal as follows.

Praxical value thesis

When an agent plays a causal role in bringing about their true belief via some intellectual performance, then

[1] the intellectual performance thereby possesses a form of instrumental praxical epistemic value, the value of causing the given true belief, and

[2] the intellectual performance also can thereby possess a form of final praxical epistemic value, a kind of eudaimonistic value.3

Here, [1] is an epistemic analog to the kind of instrumental value that obtained in the moral case, the value that X1 possesses by virtue of causing Y1. Regarding [2], the thought is that human flourishing is partly constituted by our being responsible for our true beliefs. According to Sosa,

passive reception of truth is not enough to count as human good, or at least not as the chief human good. Our preference is not just for the presence of truth, then, however it may have arrived there. We prefer truth whose presence is the work of our intellect, truth that derives from our intellectual performance. We do not want just truth that is given to us by happenstance, or by some alien agency, where we are given a belief that hits the mark of truth not through our own performance, but in a way that represents no accomplishment creditable to us (Sosa: 2003: 174).

In connection with this thought, Sosa defends a version of what is known as the Credit View of Knowledge, according to which if an agent knows that p then the agent deserves credit for believing truly that p.4 But again, this is the move that I want to leave aside here. Regardless of whether we accept the Credit View of Knowledge or Sosa’s proposed solution to the primary epistemic value problem, all my argument needs is that he has successfully put his finger on some important form of epistemic value, and that the Praxical Value Thesis is true.

I find the thesis quite plausible. In the moral case, if you push a button that saves the lives of a million innocent people, your pushing the button possesses tremendous positive moral value not because it itself was so pleasurable, but because of the pleasure it caused (and the pain it prevented). Analogously, in the epistemic case, intellectual performances that result in true beliefs possess positive epistemic value because they cause something of epistemic value. Or at least the person who denies this must explain how Sosa’s moral analogy breaks down. Because I find Sosa’s argument convincing, in what follows I will treat the Praxical Value Thesis as a premise in order to draw out further consequences that follow from it.

Turning then to those further consequences, consider three possible epistemic agents: A, A-, and A_Epi. A is an inhabitant of the actual world who holds a number of true beliefs, many of which are caused by A’s skillful intellectual performances—performances which thereby possess the two forms of praxical value described by the Praxical Value Thesis. For example, perhaps A believes truly that Idaho is contiguous with exactly six other states, and does so not as a matter of epistemic luck but thanks to their keen knowledge of American geography. For the sake of the comparisons that follow, it might be helpful to think of A as yourself.

A- is an otherworldly counterpart of A who is just slightly epistemically worse than A is. A- still holds many true beliefs, and many of these true beliefs are caused by their skillful intellectual performances, just as with A. However, A- also holds a few false beliefs on topics where A holds true ones, and A- does so as the result of mistakes in reasoning or other intellectual failures, not just bad luck. For example, perhaps A- believes falsely that Idaho is contiguous with just five other states rather than the correct answer of six, and A- holds this false belief as the result of a simple counting error, not recognizing that it is inconsistent with their individual beliefs that Idaho is contiguous with Washington, Oregon, Nevada, Utah, Wyoming, and Montana. If you think of yourself as A, then A- is a slightly epistemically degraded version of yourself. It is you-minus; you with just a few more mistakes and false beliefs.

Finally, A_Epi is an epiphenomenal counterpart of A. That is, consider a world (maybe it is an impossible world) indiscernible from the actual world in all respects except that mental states never cause any effects there, and so as a special case intellectual performances never cause true beliefs.5 We will all have epiphenomenal counterparts at this world. That is, beings physically indiscernible from ourselves, and undergoing all the same mental states we undergo, but where these mental states do not cause anything. For these epiphenomenal counterparts, you might suppose that all the real causal work is done by the brain states on which their mental states supervene.6 At this epiphenomenalist world, A_Epi will have all (or almost all) the same true beliefs that A has here in the actual world.7 For example, if A believes truly that Idaho is contiguous with exactly six other states, A_Epi will have the same true belief. If you have been thinking of yourself as A, then A_Epi is your epiphenomenal counterpart, it is what your life would be if epiphenomenalism were true.

Now, in discussing mental causation, Jerry Fodor (1989: 77) famously writes that if epiphenomenalism were true, it would be “the end of the world.” Would it epistemically be the end of the world, though—would anything of epistemic value be lost if epiphenomenalism were true? Yes, if the Praxical Value Thesis is true. In the epiphenomenal world, A_Epi’s belief that Idaho is contiguous with exactly six other states will still be true, and so will still possess the value attached to true beliefs. However, Sosa’s praxical epistemic values will be entirely missing. No intellectual performance of A_Epi will possess the sort of instrumental epistemic value described in clause [1] of the Praxical Value Thesis, since no such intellectual performance is causally responsible for any true belief. And no intellectual performance of A_Epi will possess the sort of final epistemic value described in clause [2] of the Praxical Value Thesis, the sort of eudaimonistic value that is attached to epistemic agents arriving at their true beliefs in a way that is creditable to them as agents, that is the result of their doing.

If Aristotelian flourishing consists in activity of the soul in accordance with virtue, then your epiphenomenal counterpart is forever closed off from such flourishing, on account of their completely inactive soul. What Sosa in the above passage calls the chief human epistemic good—truth whose presence is the work of our intellect—is completely unattainable for our epiphenomenal counterparts. I’m not sure if this would qualify as the Fodoristic “end of the world,” but it would be very bad. Epiphenomenalism would drain the world of an important form of epistemic value.

So then, which of these three agents would it be epistemically best to be, which should you rationally prefer to be if you were somehow given the choice: A, A-, or A_Epi? The question is meant to be easy: you should pick A. For, A scores just as high as A_Epi does with respect to the epistemic value of having true beliefs, while scoring much higher on the dimension of praxical epistemic values. And in addition, A scores at least as high as A- with respect to Sosa’s praxical epistemic values, while scoring a bit higher on the epistemic value of having true beliefs. A thus beats each of its alternatives with respect to some form of epistemic value, and never does worse than those alternatives with respect to any other form of epistemic value. So pick A.

But suppose we now take A off the table, so that the choice is between the remaining options of A- or A_Epi. Now there is a value conflict. If the only thing you care about is having true beliefs, you should prefer to be A_Epi, since A_Epi has all the true beliefs that A- does and more besides. For example, A_Epi believes truly that Idaho is contiguous with six other states, while A- believes falsely that it is five. But on the other hand, if what you care about are Sosa’s praxical epistemic values, then you should pick A-, for A- is often causally responsible for their true beliefs in a way that is to their credit, while A_Epi never is. A- is an active producer of epistemic goods, a cause of true beliefs via their intellectual performances, while A_Epi is not. A_Epi is merely a passive recipient of epistemic goods, a being that is given true beliefs thanks in part to causal processes carried out by their brain, but never by their mind, never by their intellect.

In this case, I think it is intuitively clear that you should rationally prefer to be A-. Moreover, anyone who accepts the Praxical Value Thesis or thinks that Sosa is getting at any remotely interesting form of epistemic value should agree. For, if praxical values are worth anything at all, it should not be the case that it would be epistemically better to lose them in their entirety than to lose just a few true beliefs. If you wake up tomorrow and have forgotten how many states border Idaho, so that you now falsely think it’s five, then you have lost a bit of epistemic value but it is not the end of the world. If you wake up tomorrow and your mind has turned epiphenomenal so that none of the true beliefs you acquire going forward are ever to your intellectual credit, it is an epistemic catastrophe. You have lost your chance at Aristotelian epistemic flourishing, at attaining the chief intellectual good.

Again, I don’t think we need to endorse Sosa’s view of knowledge or his solution to the primary value problem here, but framing it in those terms even just momentarily may be clarifying. On the Sosa view, epiphenomenalism would entail that we lose entirely the form of epistemic value that is distinctive of knowledge, the form of value that sets knowledge apart as something epistemically better than mere true belief. In that sense, it’s the end of the world as we know it. And I feel fine with having a preference to retain some of this distinctive value of knowledge even if it means sacrificing a few true beliefs—that is to say, I regard this preference as one that is at least rationally permissible to have.

This conclusion does not take me all the way to where I want to go, but it is an important preliminary result that already offers a kind of resonance with the IKEA effect. Here is a way to formulate this result.

Preliminary Value Proposition: It is rationally permissible for an agent to prefer on epistemic grounds alone having a belief set that contains fewer true beliefs overall but more true beliefs that involve praxical values and are to the agent’s credit.

In saying that an agent can have this preference on “epistemic grounds alone,” I mean that the preference is rationally permissible purely on the basis of the epistemic values involved, even leaving non-epistemic values aside. So for instance, it’s not just that you can prefer being A- to A_Epi on the basis that epiphenomenalism would entail no action you perform is ever to your moral credit, and this moral capacity means so much to you that you would sacrifice a few true beliefs to retain it. Rather, it’s that even if we bracket everything but epistemic value, you can rationally have this preference because of the praxical epistemic values involved.

In the passage cited above, Sosa says that “We prefer truth whose presence is the work of our intellect” over truth given to us by happenstance, and presumably he means this is a rational preference for us to have. What I am adding beyond Sosa’s claim in the Preliminary Value Proposition is that in order to satisfy this rational preference we have for truth whose presence is the work of our intellect, we can rationally sacrifice a bit of truth elsewhere in our belief set, which in effect is what you are doing if you prefer being A- to A_Epi on the basis that at least some of the true beliefs of A- are to their intellectual credit while none of the true beliefs of A_Epi are. To put the point in IKEA-friendly terms, you can prefer to have a belief set that is of your own making even if the set is of lower inherent quality (contains fewer true beliefs), in something like the way you can prefer to have an origami crane that is of your own making even if it is of lower inherent quality than one made by an origami expert.

Now, in affirming the Preliminary Value Proposition, I take no stand on just what the exchange rate should be for trading away true beliefs to obtain greater praxical value. In a footnote, Sosa (2003: 174, n. 9) mentions that Alvin Plantinga suggested to him that it might not be so bad to have a true belief that you are not causally responsible for in cases of Divine revelation, where God is entirely causally responsible for your belief instead. Maybe this is right, and maybe there are other ways in which we should be willing to give up a bit of epistemic agency in exchange for truth. I will leave these matters open here.

Lackey on testimony & credit

In Sect.2 we focused on causation as an all-or-nothing affair. Either some causal relation obtains or it does not (as in the epiphenomenal world). But there are also important distinctions to be drawn between the degrees of causal contribution made by the various causes of some effect. Some causes are more responsible than others, a point recognized by both common sense and science. For example, in a discussion of nature and nurture, even if it is agreed that both genes and the environment exert some amount of causal influence over various phenotypic traits, researchers might hold that my genes make a comparatively greater causal contribution (and so the environment less) to my trait of having two thumbs than they do to my trait of being able to read.8 Unfortunately I don’t have a rigorous account of degrees of causal contribution to offer here that might be plugged into the account of epistemic value I am developing. Instead, I will rely on intuitive judgments about degrees of causal contribution in various cases and hope that in principle such an account could be developed.

Suppose I order a preassembled bookshelf not from IKEA but from Pottery Barn, a competing home furnishing chain. I do so by going on the Pottery Barn website and clicking the “purchase” button that shows up on my computer screen. The bookshelf arrives at my house a week later, fully put together by workers at the warehouse. In this case, I still made some causal contribution to the presence of the new bookshelf in my living room. My clicking the “purchase” button was not epiphenomenal. But I did not play the same causal role as the warehouse workers who assembled it for me. I did not make as much of a causal contribution as you made to your IKEA bookshelf, what with your picking up and positioning the heavy pieces of lumber, screwing in bolts with one of those IKEA Allen wrenches, and so on.

Jennifer Lackey (2007) presents a case that I want to treat as an epistemic analog to this, a case involving testimony. Suppose that upon arriving at the Chicago train station, Morris wants directions to the building formerly known as the Sears Tower.9 He asks the first passerby he sees, someone who happens to be a Chicago resident with extensive knowledge of the city. The passerby gives Morris the correct directions: the Sears Tower is two blocks to the east of the train station. By accepting this testimony. Morris forms a true belief about where the Sears Tower is, a true belief that I will assume rises to the level of knowledge.10

Lackey uses the case to press an objection against the Credit View of Knowledge that we mentioned in passing in Sect.2, the view held by Sosa and several other virtue epistemologists. Her thought is that Morris knows where the Sears Tower is but does not deserve credit for this intellectual achievement—rather, the credit goes to the passerby with the detailed knowledge of Chicago geography, the one who gave Morris directions. If Lackey is right, the case is a counterexample to the claim that if an agent knows that p, then the agent deserves credit for believing truly that p. But just as we remained neutral on Sosa’s defense of the Credit View, I want to remain neutral on Lackey’s critique of it. For present purposes, all that I need is that Lackey is getting at something correct about epistemic value.

Here is how I understand the correct point she is getting at, formulated as a modification of Sect.2’s Praxical Value Thesis.

Degrees of praxical value thesis

When an agent plays a causal role in bringing about their true belief via some intellectual performance, then

[1] the intellectual performance thereby possesses a greater degree of instrumental praxical epistemic value, the greater extent to which the intellectual performance is causally responsible for the true belief, and

[2] the intellectual performance also can thereby possess a greater degree of final praxical epistemic value, a kind of eudaimonistic value, the greater extent to which the intellectual performance is causally responsible for the true belief.11

Perhaps Morris deserves some credit for his newly acquired true belief about where the Sears Tower is located. Sosa (2007: 95), in considering cases of testimony, suggests that someone like Morris might deserve partial credit.12 But not too much. And in particular, Morris does not deserve as much credit for his true belief about where the Sears Tower is located as the Chicagoan passerby deserves for their own true belief about the matter. In connection, the intellectual performance of the passerby plausibly possesses a greater amount of Sosa’s praxical values than Morris’s intellectual performance does. The passerby gets full credit; Morris gets just partial credit.

Lackey (2007: 358), following Greco (2003: 131), frames her discussion in terms of “causal salience”: the passerby’s intellectual performance is the more salient part of the causal explanation for how Morris gets his true belief than Morris’s own intellectual performance is. But I take that to be just another way of getting at the thought that I have been expressing in terms of “degrees of causal contribution”: the passerby’s intellectual performance makes more of a causal contribution to Morris’s true belief than Morris’s own intellectual performance does. Morris merely pressed a kind of intellectual “purchase” button in his mind when he approached the passerby; it was the passerby who preassembled the Sears Tower belief for him.

I find the Degrees of Praxical Value Thesis intuitively compelling. In addition, if we continue to accept the original Praxical Value Thesis defended by Sosa—and as I stated back in Sect.2, I am content to treat that thesis as a premise in the present paper—it is hard to see a promising way to resist this modified version of it. If Sosa is right and there is something of value in truth whose presence is the “work of our intellect,” then isn’t it better when it is squarely the work of our intellect rather than merely tangentially so? If part of what we want is to live lives in which we are active producers of epistemic value rather than just passive recipients of such value, then surely we should want more than to make just the most minimal causal contribution imaginable to our true beliefs, to just barely cross the threshold above epiphenomenalism.

In other domains of value, we seem to accept such a claim about degrees. Imagine we are at the house of our friends Katherine and Michael, a pair of amateur carpenters, and you are praising the delightful mahogany coffee table that they have handcrafted, giving them a lot of credit for their work. It would be out of place for me to interrupt and insist that I deserve equal credit because I have recently purchased from Pottery Barn a coffee table that is at least as good as Katherine and Michael’s table. Katherine and Michael are the creators of coffee table-value in a way that I am not, given the vastly different types and degrees of causal contribution we have made. As it goes with coffee tables, it also goes with true beliefs, if the Degrees of Praxical Value Thesis is correct.

Of course, it would be absurd to pursue greater degrees of praxical epistemic value monomaniacally, refusing to ask for directions to the Sears Tower (or anywhere else), refusing to rely on testimony from anyone on any topic. Our knowledge is heavily, heavily indebted to the intellectual achievements and testimony of others, and we would sacrifice a tremendous amount of epistemic value in our own lives—so many true beliefs, so much knowledge—if we refused to make use of this resource. It would be like living a life in which you committed to building your own house entirely by yourself and from scratch, growing and preparing all of your own food, making all your own tools, and on and on. We are vastly epistemically better off thanks to the division of intellectual labor (Kitcher, 1990), just as we are vastly materially better off thanks to the division of manual labor.

But we can agree to this point while holding that something important of epistemic value would be lost in a life in which you never make much of a causal contribution of your own to your true beliefs. To crystallize the point, let us return to the type of argument presented in Sect. 2. Consider three possible epistemic agents: A, A+ , and A_lackey. A is, once again, some ordinary agent here in the actual world, someone who holds various true beliefs and often does so as a result of their own intellectual achievements, so that they often possess Sosa’s praxical values in a high degree. Again, you might think of A as yourself. Or, better, so that I don’t give offense when I speak frankly about A’s limitations in what follows, you might think of A as me.

Let A+ be an epistemically upgraded version of A inhabiting another possible world. Suppose that A+ has every true belief that A does and more besides, while often avoiding A’s false beliefs. In addition, A+ has more and greater intellectual achievements whereby they arrive at true beliefs through their own intellectual performances. For the sake of the present argument, we can even let A+ be a significantly upgraded version of A. For example, you might think of A+ as an Einstein-like figure, with accomplishments similar to the development of the theory of relativity.

Finally, let A_lackey be a worldmate of A + , and let them be an intellectual “lackey” (pun intended) to this Einstein-like figure. What I mean is, A_lackey holds all the same true belief that A+ holds, but does so in all cases as a causal result of A+ first figuring out these truths by themself and then passing the truths along to A_lackey via testimony, in exactly the way the Chicagoan passerby gave Morris directions to the Sears Tower. Accordingly, A_lackey never holds any true beliefs as the result of their own intellectual accomplishments, or at least never as the result of what are mostly their own accomplishments (to allow that A_lackey might still deserve some minimal, Morris-like degree of partial credit).

It may be that there are other forms of epistemic value that our discussion has overlooked so far. If so, I want to build into the case that these other forms are fully present in A_lackey’s case, at least to the extent that they are independent of Sosa’s praxical values. So, for example, John Kvanvig (2003) draws attention to the epistemic value of understanding as distinct from that of knowledge or true belief. You could worry that this value will be missing in A_lackey, especially if you imagine them as merely mouthing the various truths that A+ passes along, affirming their content without really thinking them through. “Spacetime is curved,” says A_lackey unthinkingly, not really knowing what it means, but believing the sentence is true because it is what A+ said. But we can rule this out by stipulation. Suppose that in the case, A_lackey comes to understand the various truths A+ passes along just as well as A+ does, thanks to A+ ’s patient explanations. “Spacetime is curved,” says A_lackey knowingly, understandingly.

If you were given a choice to be one of these three agents, it is epistemically best to be A. If tonight I go to sleep as A, and overnight something happens to me so that I wake up as A+, with Einstein-like intellectual abilities and capacities, and intellectual achievements going forward that are comparable to Einstein’s Annus mirabilis papers of 1905, that would not be a tale of epistemic tragedy. It would be a wonderous epistemic miracle. It would be me nudging upwards my epistemic grade in life, going from an A (being generous) to an A+.

But suppose we now take A+ off the table, so that the choice is between the remaining options of A or A_lackey. Then, if we focus just on the value of true belief, A_lackey scores more highly than A does—for again, A_lackey is the match of A+ with respect to true beliefs, and therefore is superior to A. However, if we focus instead on degrees of praxical epistemic values, it is A who scores more highly—A possesses such praxical epistemic values in relative abundance, while A_lackey does not. So once again, we find ourselves with a conflict of epistemic values.

Given this choice, I pick A. For again, I want to live a life in which I am an active creator of epistemic goods, not just the passive recipient of epistemic goods preassembled by others. I want to make a substantial causal contribution of my own to the advancement of knowledge, or even just to my own knowledge or true beliefs. If the Degrees of Praxical Value Thesis is true, there is a form of epistemic value that can be obtained only by thinking for yourself, in the sense of making some sort of substantial causal contribution to your true beliefs, beyond the minimal type of contribution that Morris or A_lackey make. Thinking for yourself in this sense is, I suggest, a component of Aristotelian flourishing, so that a life in which all your true beliefs are largely causally dependent on the intellectual labor of others, as we find in the case of A_lackey, is a life closed off at least to some extent from the chief human good. It is a life in which you can obtain partial flourishing credit, at best.13

Allan Hazlett (2016) argues that non-deferential belief has a certain distinctive value, but it is a social value. For one thing, Hazlett contends, non-deferential beliefs can contribute to a form of collective reliability, at least in certain conditions, making your social group more likely to settle on true beliefs thanks to the “wisdom of crowds” (Surowiecki, 2004), while deferential beliefs fail to promote such reliability because they are not independent of the beliefs of other members of the group. For another thing, Hazlett adds, non-deferential beliefs can contribute to the democratic legitimacy of your social group, again at least in certain conditions. Here the thought is that genuine collective deliberation requires the members of your group not simply defer to the judgment of some dictator or a narrow class of technocrats about what to do as a group, but instead to engage in public deliberation where different individuals bring their own beliefs to bear. Hazlett’s view potentially could be used to try to undermine my argument here by maintaining that the sort of value I am identifying in cases of non-deferential belief is not epistemic value at all but only social.14

In response, I want to appeal to a thought experiment introduced by Charles Taliaferro (1985) and later used by Linda Zagzebski (1996: 24–28). Suppose there is God, who knows all truths through the exercise of his own cognitive power. And then there is Claude, a confidante of God’s who also knows all truths but only because God first figures them out and then passes them along to Claude via testimony.15 In this scenario, both Taliaferro and Zagzebski maintain, God is epistemically superior to Claude even though Claude knows all truths.16 I share this intuition, and understand it in terms of degrees of praxical epistemic value. God’s intellectual achievements possess a greater degree of praxical epistemic value than Claude’s do because God is a producer of epistemic goods in a way Claude is not.

The case is useful for my purposes because it cannot readily be understood in terms of Hazlett’s social values. It is not as if the group that consists of God and Claude could somehow improve their collective reliability if Claude were less deferential. After all, God is unerring, and so a policy by which the group always defers to God cannot be beaten in terms of reliably settling on true beliefs. And it is not as if the notion of democratic legitimacy has much purchase on a group that includes God as a member. What, is the majority going to outvote God, with God then expected to accept their decisions based on democratic principles of how free and equal citizens should engage one another in matters of public deliberation? This feels like the premise of a joke.

No, social values aside, God is epistemically superior to Claude because he is the active producer of epistemic goods while Claude is merely a passive recipient of the goods that God makes. But, once it is granted that God is Claude’s epistemic superior, my argument has all the room it needs to maneuver. For if God is epistemically better off than Claude, there is room for an epistemic value conflict between truth and that dimension of value that makes God epistemically superior, and for scenarios in which it is rationally permissible to trade off a bit of truth in exchange for this other epistemic value.

Back near the end of Sect.2, I set out the Preliminary Value Proposition. Now, as we approach the end of Sect.3, I can offer a refined version of the thesis that takes into account degrees of praxical value.

IKEA Value Proposition: It is rationally permissible for an agent to prefer on epistemic grounds alone having a belief set that contains fewer true beliefs overall but more true beliefs that the agent made a greater causal contribution to, and so that involve a greater degree of praxical values and are more to the agent’s credit.

What this adds to the Preliminary Value Proposition is the claim that epistemic agents can rationally prefer not just to make some causal contribution, however slight, to their true beliefs, but in addition that they make a greater causal contribution. We don’t just want to pass the minimal threshold of not being epiphenomenal, we want to play a substantial causal role in producing epistemic goods like true belief, as opposed to the role of mostly just being the passive recipients of epistemic goods manufactured by others.

It may be helpful to approach the IKEA Value Proposition from another angle, so consider a further line of argument. Like any fallible being, I am bound to have some false beliefs in my belief set in addition to many true ones. In addition, I have some beliefs that I have made a substantial causal contribution to—beliefs I have thought long and hard about, published papers about, and so on—and others that I have adopted from my community without much intellectual labor of my own, Morris-style. Now, suppose that these two ways of dividing my belief set match up in the following way. Every belief that I have made my own substantial causal contribution to turns out to be false, and so my true beliefs are exclusively those that I have adopted from others. Every time I have personally intellectually tried in my life, my intellectual performance has been a failure (where success is measured in truth). But thankfully, despite this unrelenting string of personal intellectual failures, I have been able to rely on the successful intellectual achievements of others to get by relatively well, to find my way to the Sears Tower and whatnot.

This does not sound like a recipe for human flourishing! I do not want to live a life of unrelenting personal failure but buoyed along by my more successful acquaintances. Even if it allows for some modicum of eudaimonistic value—again, maybe I can earn partial eudaimonistic credit—it seems plausible that I would achieve such value in a greater degree if at least some of my own intellectual endeavors were successful, that is, if some of the beliefs where I made a greater causal contribution ended up being true. If the IKEA Value Proposition is true, then other things being equal, I can rationally prefer that if I am going to have any false beliefs at all, that they not be distributed so as to minimize my possession of praxical epistemic values. Avoiding such a distribution would allow me to better tap into the distinctive value of being an active producer of epistemic goods.

Now, the IKEA Value Proposition is a philosophical thesis about epistemic value. When Stafford (2021) suggests that there may be an epistemic IKEA effect, in contrast, he is entertaining a psychological hypothesis about how human minds work. In the next, final section of this paper, I turn to consider how the IKEA Value Proposition might bear on the target of Stafford’s discussion: conspiracy theorists. In the process, I don’t mean to commit myself to any particular psychological account of what drives conspiracy theorists, or to any proposal about what should be done about them. Rather, the idea is to use the case of conspiracy theorists to say more about the sort of epistemic value conflicts that I have focused on in making my arguments in Sect.2–3.

Conspiracy theorists

The world is in a bad way, and at least part of the problem is epistemic. There is a disconcertingly widespread embrace of implausible conspiracy theories that are at odds with our best evidence and understanding of how things work. Theories about who was responsible for the JFK assassination or for 9/11, theories about how scientists are plotting to propagate a myth of global climate change, theories about the risks of vaccines, and much more. Accordingly, conspiracy theorists have been a target of much recent empirical work by psychologists, political scientists, and others aiming to understand the causal processes involved.17 But also, and to some extent in connection, conspiracy theorists have increasingly become a target of interest to epistemologists.18

Here is one line of thought that a virtue epistemologist might have. The embrace of conspiracy theories often reflects some kind of epistemic vice among the theorists.19 There are different candidate vices you might point to here, but one we might focus on is that of intellectual arrogance.20 To pursue this route would be to hold that conspiracy theorists are often unjustifiably self-confident in their knowledge and understanding of the world, and too quick to discount the views of others, especially those of experts. “Jet fuel can’t melt steel beams,” I continue to assert dogmatically even after you have patiently explained for the umpteenth time how my views about 9/11 have been debunked by people who know better.

If intellectual arrogance is the problem, then an obvious solution is to try to promote intellectual humility, understood roughly as the epistemic virtue of recognizing and attending to one’s own intellectual limitations.21 If only I could somehow be instilled with such humility, I would lose confidence in my 9/11 theories, clearing the way for my reception of expert opinion on the topic. There are epistemic goods out there to be had, in the form of true beliefs produced by experts, but my current intellectual arrogance gets in the way of my receiving those goods.

There are different ways you might go about trying to instill greater intellectual humility in people, but to have a concrete proposal on the table, consider a suggestion by psychologists Steven Sloman and Philip Fernbach (2017: 191). They envision promoting intellectual humility by exploiting the illusion of explanatory depth discovered by Leonid Rozenblit and Frank Keil (2002). People often think they know how various everyday objects work in detail, but this is an illusion that is punctured when you ask them to provide a step-by-step explanation. For example, ask people how well they understand how a toilet flushes or how a zipper zips and they will generally say quite well. But ask them next to give a detailed, step-by-step explanation of the mechanisms involved and they come to recognize they cannot. This characteristically induces a kind of humility—people revise their original beliefs and acknowledge that they do not understand such objects after all. I started off with a haughty zipper arrogance, boasting to anyone who would listen how well I understood zippers, but in the end it was replaced with a chastened zipper humility.

Sloman and Fernbach (2017: Ch. 9) observe that the same dynamic has been shown to play out on contentious political topics, and suggest that we might reduce people’s confidence in extremist views and increase their intellectual humility simply by asking them to provide detailed explanations on such matters. Ask them to explain in detail how a single-payer health system works, or how steel beams do in fact melt, and when they cannot they might just become less confident in their views. This clears the path for people to rely more on experts than they presently do, a result that Sloman and Fernbach (2017: 188) recognize has an elitist air about it and stands in a certain tension with democratic ideals, but also one that tends to deliver more true beliefs.

But now, here is a second, distinct line of thought that an epistemologist might have about conspiracy theorists. As expressed by the IKEA Value Proposition, there really is a genuine form of epistemic value to be had by making your own substantial causal contribution to your true beliefs, a form of value that cannot be obtained by deferring to others, even experts. It is the value of thinking for yourself, of being an active producer of epistemic goods, and not just a passive recipient or consumer of such goods. When you try to instill greater intellectual humility in people, whether this is by asking them to explain zippers or through some other means, you almost inevitably ask them to forego some amount of this epistemic value, notwithstanding that such humility might be truth-conducive.

Now, the sort of epistemic praxical value that has been my focus throughout the paper does not obtain in cases in which conspiracy theorists hold false beliefs. For, following Sosa, I have taken such praxical value to be attached to causing true beliefs, and so where there is no true belief as the end product, there is no such epistemic praxical value. And so the idea here is not that conspiracy theorists might weigh the option of deferring to experts and obtaining the truth versus thinking for themselves and obtaining the false, and then rationally prefer the latter, taking it to possess greater overall epistemic value. Rather the thought is that deferring to experts inevitably involves forgoing epistemic praxical value to any great degree, while thinking for yourself offers at least the prospect of obtaining this value, of enjoying an intellectual achievement greater than deference to experts can offer, provided you can successfully think your way to the truth.

To be sure, conspiracy theorists often err by overestimating the inherent quality of the belief sets they produce, in something like the way that subjects in the IKEA effect experiments run by Norton and colleagues overestimate the inherent quality of the origami frogs and cranes they made. That is, they regard the belief sets they have built as superior in inherent quality to those built by experts—superior in the sense of containing more truth or likelihood of truth—and perhaps this is the result of the intellectual labor they have poured into them. This is a mistake, a cognitive error. So, my claim is not that conspiracy theorists are exhibiting a rationally permissible preference. Rather, my claim is that even if we understand them as thus proceeding irrationally, perhaps there is a way to allow them to pursue the value attached to being a producer of epistemic goods while nudging them along the path to truth, giving them a chance to think for themselves.

Stafford (2021) reports that in a digital ethnography of vaccine-hesitant parents, a pattern that stands out is that it is common for people who believe in vaccine myths (e.g., that they cause autism) to emphasize how much research they have personally done, how many hours they have spent reading studies for instance. “Do your own research” has emerged as a kind of slogan for QAnon conspiracy theorists (LaFrance, 2020).22 Rather than construing this purely as a matter of intellectual arrogance—who in the world are these people to think they are qualified to judge the relevant research?—you might think it reflects at least in part a legitimate desire to pour their own intellectual labor into something important in their lives. In that case, the remedy to pursue would not be to try to embarrass or browbeat them into greater humility so that they can become better passive recipients of the epistemic goods produced by experts. Instead, you might try to rechannel their active epistemic agency in other directions.

There are different forms this could take, but one option that Stafford mentions is represented in work by cognitive scientists Sacha Altay et al. (2021), showing that having people actively engage with a chatbot in possession of information about COVID-19 vaccines was an effective way of steering them toward positive attitudes about the vaccines, in comparison with a control condition in which people passively received a summary of vaccine information. Stafford takes this finding to suggest an epistemic IKEA effect, with the idea being (at least as I would put it) that the chatbot gives people room to exercise epistemic agency, trying out objections and otherwise constructing their beliefs through their own intellectual labor, in a way that receiving the preassembled summary of information does not.

Given the spirit of my discussion here, I again don’t want to overcommit to this proposal. Instead, the more general point is that if the IKEA Value Proposition is true, then we should expect rational epistemic agents to want to have opportunities to demonstrate their intellectual abilities and skills or even mastery, to pursue intellectual achievements that allow them to be producers of epistemic goods. Approaches that neglect epistemic agency and focus instead just on how to maximize the passive reception of expert opinion or other forms of epistemic goods made by others are bound to miss this. Even as our epistemic communities develop better ways to deliver intellectual cake mix to people, we might want to allow them room to contribute the labor of adding their own intellectual egg.