1 Why robust virtue epistemology has a problem with environmental luck

Gettier cases are usually described as situations where some justified true belief (JTB) does not count as knowledge because it is lucky in some particular way (Gettier 1963). Here is a famous case (adapted from Chisholm 1966, p. 23, fn22):

SHEEP: Nora comes to believe there is a sheep in a field by looking at something that very much looks like a sheep. What Nora sees is in fact just a shaggy dog but, by pure chance, behind that dog is a sheep that she cannot see.

Although Nora’s belief would be true, and she seems to be well justified in believing it, her justified belief would be true by pure chance. Most epistemologists intuitively claim that it falls short of knowledge.

Virtue epistemology, in its different varieties, is a prominent way of accounting for these cases by appealing to the etiology of the JTB —in particular, to the way it was acquired by the agent. Let me consider just two of its varieties here: according to Ernest Sosa’s version (2007), (2009), (2011), the agent’s attaining her JTB must be a manifestation of the relevant cognitive abilities, in the sense that her abilities are not only what explains why she has that particular belief, but also why the content of her belief is correct. In the case of SHEEP, Nora attains her JTB thanks to her abilities, but the truth of her belief does not manifest those abilities at all. According to John Greco’s version of virtue epistemology (2010), (2012), the correctness of the belief must be due to the faculties and virtues of the agent in such a way that she deserves credit for it, and Greco adduces as a reason for this that knowledge is a social label we employ to flag firmly reliable sources of information. Since Nora got it right by pure chance, she doesn’t deserve credit for the correctness of her JTB, and that is why, according to Greco, we don’t call it knowledge.

Sosa and Greco share the idea that knowledge essentially consists in some kind of achievement of the agent: a specific kind of success (getting it right) that is due to the agent’s cognitive abilities. Theirs are two varieties of robust virtue epistemology because they claim that by adding to the classic three requirements (JTB) the further condition of representing some genuine cognitive achievement we may be on the right track to finding out a set of necessary and sufficient conditions for knowledge—or at least we would not need to add any further condition to deal with cases of knowledge-undermining luckFootnote 1. The robustness of those virtue epistemological accounts consists in their claim that knowledge may be defined as JTB cognitively achieved in the right way, whose truth is a result of the agent’s faculties, virtues and abilities.

But not all virtue epistemologists are happy to be so robust. Pritchard in particular has claimed that such robust accounts cannot deal with different objections that, taken together, he considers devastating (2010a, pp. 25–47, 2010b, pp. 27–29, 2012, pp. 264–271). Two of those objections allegedly show that knowledge and achievements may fall apart. That is, we may find counterexamples to robust virtue epistemology where an agent has some knowledge whose truth she did not achieve, and we may also find cases where, despite the fact that an agent does achieve some truth, her JTB falls short of knowledge.

The former possibility is exemplified by cases of easily acquired testimony—see Pritchard (2012, p. 269), based on Lackey (2006). In my opinion, some convincing answers have already been given to this objection—answers that first highlight the specific abilities that are required in order to find trustworthy sources of information even in very favourable situations, and then complement this idea with some social conception of distributed knowledgeFootnote 2. For this reason, in this paper I will focus on the latter objection, which goes in the opposite direction, purporting to show that we may find JTBs cognitively achieved that fall short of knowledge. Such a possibility is allegedly exemplified by Alvin Goldman (1976) in his famous fake barn case:

BARN: Barney comes to believe there is a barn in a field by looking at something that very much looks like a barn. Unbeknownst to him, he is in a county full of fake barn façades that look exactly like real barns from that distance. Fortunately, the barn he is looking at is one of the very few real barns there are in that county.

Barney is forming a true belief and he seems to be well justified in believing it, but he is very lucky that his belief is true. It could have gone wrong quite easily, and he would still have formed that belief in the very same way. For that reason, most epistemologists claim that Barney’s JTB, just like Nora’s, falls short of knowledge.

Although SHEEP and BARN share one same diagnosis their respective explanations are importantly different. We could say that Nora was lucky to have formed a JTB, but she did not achieve its truth, whereas Barney was lucky to achieve his JTB. In other words: while Nora’s belief in SHEEP is successful because of luck (since luck is the reason why her belief is true), the truth of Barney’s belief seems to be attained by him as a successful result of the exercise of his cognitive abilities, despite the fact that he was very lucky not to fail in such an epistemically inhospitable environment. Nora simply failed (she mistook a dog for a sheep), and then luck intervened to make her belief true, whereas Barney succeeded (he took a barn for what it was), although he could have easily failed. Luck was lurking in Barney’s environment, but it did not intervene, and this is why Pritchard (2010a, p. 36, 2012, p. 267) insightfully labelled that kind of luck as “environmental luck”, in contrast to the one affecting classic Gettier cases, such as SHEEP, which he labels as “intervening luck”.

Pritchard is definitely right in claiming that cases of environmental luck are a much tougher nut to crack for robust virtue epistemology than cases of intervening luck. The reason is that, while intervening luck undermines both knowledge and achievement at the same time, environmental luck seems only to undermine knowledge, while it apparently preserves the cognitive achievement. Robust virtue epistemologists seem thus to have a problem with Barney, since his case apparently shows that knowledge and achievement may come apart, while those accounts would justify the opposite prediction.

At this point, those virtue epistemologists who want to remain robust have to face an unpalatable dilemma. Its first horn is to deny our intuition in BARN by claiming that there actuallyis knowledge in environmental luck cases, precisely because there is some sort of achievement, but they have to hold at the same time that Barney’s knowledge is, so to speak, second-class. Ernest Sosa (2007, pp. 31, 92–112), (2011) has decided to go for this horn, by holding that agents affected by environmental luck may still have what he calls “animal knowledge”, while they would lack “reflective knowledge”, which is the agent’s awareness of her own reliability in that epistemic environment. And, by the same token, they would also lack “full knowledge”, understood as the higher degree of knowledge that is conferred to the agent’s belief when it is a manifestation of her reflective knowledgeFootnote 3.

Those robust virtue epistemologists that are not happy with this horn of the dilemma may choose to be impaled by the other one, which is to load their accounts up with some further explanation in order to deal with environmental luck cases. This is the solution offered by Greco (2007, p. 67), which adds to robust virtue epistemology the possibility of defining the relevant abilities in extremely fine-grained ways. Applied to BARN, Greco’s idea is that, by stipulation, Barney does not have the ability ‘to discern real barns from fake ones just by looking at their façades’, and for that reason his cognitive success may not represent an effect of the relevant ability. Since Barney does not have that accurately defined ability, Greco denies that his cognitive performance may be properly considered as his genuine achievement.

Pritchard finds both options unconvincing. He considers that Sosa’s move leads to quite a counter-intuitive claim, by attributing some sort of knowledge to Barney, even if it is of the animal kind (Pritchard 2012, p. 267, 271n38). And he sees Greco’s movement as ad hoc, and dangerously revisionary: an artificial way to define abilities alien to our common practices in ordinary life (2012, p. 268). Discarding those robust alternatives, Pritchard’s positive proposal is to preserve the main insight of virtue epistemology (i.e., that knowledge must be the effect of the agent’s cognitive abilities) while at the same time weakening the requirements it imposes on knowledge (i.e., the truth of the JTB does not have to be due to the agent’s performance). An agent must form her JTBs based on her own abilities, but she does not have to achieve its truth herself. Pritchard also proposes to complement this modest version of virtue epistemology with some sort of anti-luck clause, which would exclude merely lucky guesses. In particular, he favours a safety clause, considered as an independent and complementary requirement to the ability clause of his modest virtue epistemology. According to Pritchard’s account, knowledge would be JTB, which is an effect of the cognitive abilities of the agent, and whose truth is safe.

Pritchard’s anti-luck modest virtue epistemology does not identify knowledge with the cognitive achievement itself, and thanks to that it successfully predicts our intuitions in situations where knowledge and achievements seem to fall apart. In the case of easy testimony, the agent gains a JTB thanks to her own abilities and virtues (in particular, she is able to find good sources of information in her environment), but she does not achieve the truth of her JTB, which is mostly creditable to her informant. And in cases of environmental luck, such as BARN, the agent achieves some JTB, and he is creditable for its truth, but his achievement is not safe, and that would be the reason why it does not constitute knowledge.

My aim in this paper is not to evaluate Pritchard’s positive account, his modest anti-luck version of virtue epistemology, but only to confront the second part of his negative argument: the one where he criticizes robust virtue epistemologies in the light of cases of environmental luck. I intend to offer a new way to deal with those cases that preserves the common core of robust virtue epistemological accounts, remaining neutral on their internal disputes. As I will later show, my proposal also differs from those recently offered by Carter (2011), Carter (2014), Jarvis (2013) or Littlejohn (2014) in that it is based on the idea that we may only consider as achievements those objects, states or events that result from intentional action, and we may only extend those attributions to descriptions of those performances and their effects that may still be considered intentionally pursued by their agents. In this way, I hope to both reinforce robust virtue epistemological accounts and propose some insights that may help develop it into a fully agential theory of knowledge.

The structure of the paper is as follows: in Sect. 2, I will introduce some basic ideas about human agency, namely that actions may only be considered as intentional under some descriptions but not others, and that this depends in part on the awareness the agent has of what she is doing (i.e., on her practical knowledge); I will then make a more controversial claim, which is crucial for my argument: that only intentional actions and their intended effects may be properly considered as achievements. In Sect. 3, I will apply this idea to cognitive performances affected by environmental luck in order to show that, insofar as the agent was not aware that she was acting in that specific epistemic environment, her cognitive performance may not constitute a genuine achievement under that description. I will show then why this idea would allow us to account for this sort of knowledge-undermining luck from within the framework of robust virtue epistemology in a brand new way. And finally, Sect. 4 deals with some possible objections to my argument, most of which are attempts to undermine the claim that achievements are limited by the scope of intentional action.

2 Reflections on agency: an intentional condition for achievements

I will take it as a given that we only call “achievements” objects, states or events that result from somebody’s action. Niagara Falls, for instance, or the Amazon rainforest, are certainly amazing, but they are not achievements, in contrast to the Egyptian pyramids or the Eiffel tower. Achievements must be made by someone, they don’t just happen.Footnote 4

However, not all of our actions may count as our achievements—despite the way some people fill out their CVs! For sure, we did not achieve those things we did not succeed in doing (although some failures do often involve related achievements). Achievements imply success. But even in cases where we do succeed, our success may still not represent an achievement. For instance: the child that brings all his homework to school which has been done by his parents may pass his course, and he may thus succeed in it, but he did not achieve it. And if he passes a multiple-choice test by sheer luck, he would succeed in it, but he would not have achieved his success either. For reasons like this Pritchard (2010b, pp. 19–22) rightly postulates that there are at least two necessary conditions for achievements: the performance must count as successful, and this success must be attributable to the exercise of the relevant ability of the performer in a significant way, rather than to some external factor, such as luck, or the illegitimate intervention of other agents.

Taking a closer look at this issue, Pritchard considers that such a simple analysis of achievements ought to be refused, since it fails to account for what he calls the problem of easy achievements (2010a, p. 23, 2010b, p. 29). In normal situations, he argues, extremely easy performances, such as raising one’s arm or blinking one’s eyes, do not seem to represent achievements, even if we have those basic abilities and we do succeed in manifesting them. Pritchard claims that achievements seem to require difficulty, overcoming obstacles, which is part of their distinctive value. However, Pritchard himself recognizes that there are cases of successful performances that are quite easily attained by their agents, but still do count as genuine achievements. This is the case of excellent performances attained by agents who are extremely proficient in the task, such as Tiger Woods sinking a putt with ease, or Rafael Nadal winning a shot with no trouble at all. Trying to overcome the problem of easy achievements while still preserving the intuition that those cases do represent achievements, Pritchard proposes a bifurcated analysis of the concept: in order to represent an achievement, success may either be due to the exercise of abilities to a particularly significant level (as in the case of masters attaining achievements that are easy for them) or to the exercise of abilities which involve overcoming a significant obstacle.

I am reluctant to follow Pritchard in this bifurcated analysis. My reason for this is that such a starting point would trivially leave robust virtue epistemology out of the market, since cases of easy knowledge, which are probablly most of what we know in our ordinary lives, do not seem to fit into any of Pritchard’s alternatives. I know it is a sunny day while walking outside in the sun; Carol knows where she left her car two minutes ago; Mike knew her mother would visit him today because she told him so. It is hard to deny that those are genuine cases of knowledge, while it seems that forming those beliefs was not harder for their agents than rising their arms or blinking their eyes. They did not exercise their cognitive abilities to any particularly significant level, nor did they overcome any significant obstacle. Pace Pritchard, I think that we just have two options here: either we accept that there are easy achievements, or we reject that there is easy knowledge. Since the latter does not seem to be a sensible move, I decidedly go for the former, and will employ a notion of achievement here that does not require overcoming difficult obstacles at allFootnote 5.

That leaves us again with our two basic conditions: success and ability. They are closely related to each other up to the point that perhaps we should not even say that there are two conditions, success and ability, but only a single, complex one: success because of ability—which will be one of the basic tenets of virtue epistemology. But even in this formulation, there is doubtlessly much more to be said about achievements in order to fully understand their conceptual nature. In particular, it is crucial for my argument to point out one more idea, which is what I will call the Intentional Condition for Achievements (ICA):

ICA: Only intentional actions and their intended effects may count as achievements.

One may defend ICA as an independent third condition, together with success and manifestation of ability, although it is also possible to opt for a simpler account by including intention as an implicit requirement of the other two. This would be my preferred way of introducing intention into the picture for at least two reasons. The first one is that we could hardly consider as successful some effect of an agent’s performance that was not pursued, or at least envisaged by her at all. Successful actions are defined by the agent’s goals, and not just by her results. They thus imply certain positive motivational states on the part of the agent, since she must actively pursue the results of her performance. Accepting an impoverished notion of success, completely independent of the agent’s intention, would force us to claim that agents are ‘succeeding’ all the time in an infinite number of actions they are not even aware of, which seems quite an outlandish conclusion. So, in my opinion, genuine success already implies intention. That is why, in what follows, I will always put scare quotes around ‘success’ when this condition is not met.

But there is a second and more important reason for ICA: intention seems to be already implied by the ability condition for achievements, since only those performances that are intentional actions may count as genuine manifestations of abilities. Abilities are not just any dispositional state of ours, but only dispositions that are, so to speak, at our disposal, and are thus our powers. We do not consider mechanical and uncontrolled tics of a person as her ‘abilities’ if they are beyond her control, and we do not consider our heart pumping as an ability of ours, even if it is something that reliably happens in our bodies. Abilities are dispositional states attributed to people at the personal level, and they define a range of possible intentional actions, not just of reliable occurrencesFootnote 6.

Therefore, I think that, by assuming the two basic conditions for achievements—i.e. success and ability—, we have already brought intention into the picture, which is my preferred rationale for ICA. However, my guess is, if ICA seemed unsatisfactory, we would have to find the way intention fits into our account of achievements. Even if we opted for impoverished notions of ‘success’ and ‘ability’, intention would still have to come into the picture at some point or another, because it does not seem that we could make attributions of achievements to performances that are totally deprived of intentional characterFootnote 7.

If ICA is right, achievements are a species of intentional performances (or their intended effects) and, for that reason, they are constrained to intensional contexts. This is nothing but a platitude on intentional actions according to the mainstream today on the topic: we may consider an event as an intentional action only under some descriptions, but not under some other ones (Davidson 1980): the consideration of an event and its results as an intentional (with a t) action is limited to an intensional (with an s) context, which means that it fails to pass extensionality tests. For example, from the facts that John flirted with Kathy intentionally, and Kathy is the wife of John’s boss, it does not necessarily follow that John flirted with his own boss’s wife intentionally. The scope of the intentional action does not extend to descriptions of the action that were not what the agent was intending to do, even if those other descriptions are extensionally equivalent. The agent’s awareness of what she is doing constrains our possible consideration of an event as her intentional performance, and thus our possibility to legitimately praise its different results as her achievements.

Another way of stating this is by claiming that intentional action requires awareness, or even knowledge of what one is doing (which is called, at least since Anscombe (1957), “practical knowledge”)Footnote 8. Descriptions of the action the agent is not aware of doing would not count as her intentional performances, given that the scope of intentional action is more limited than the one of conscious action. Agents may be aware that they are performing an action they are not intentionally doing (just as when someone realizes that she is nervously tapping on the table, or that she is keeping the sun off somebody else’s eyes).

On reconsideration, it could be argued that the requisite of knowing what one is actually doing is too strong: it could be enough that one knows what one is trying to do. I will say more on this in the final section, but for now it is enough for my argument to claim that an agent may not be said to have done something intentionally unless she knew, or would later know by reflection, that it was that what she was intending to do. And, by the same token, she may not be said to have achieved it.

An important caveat before going on: some readers could be reluctant to assume ICA because they have a too demanding notion of intention in mind. One could think, for instance, that intentions are occurrent psychological states, previous or simultaneous to action, consisting on explicit representations of the intended performance. Those having this notion of intention would probably believe, quite appropriately, that ICA is too demanding, because it over intellectualizes performances, and is in conflict with the phenomenological datum concerning skilled performances, which are usually impelled by this kind of explicit considerations (see Dreyfuss 2007, discussed in Stanley 2011, pp. 23–24 and 167–184). However, the idea of intention I am employing here is much weaker than that. It is just a matter of the performance making sense in the overall conception the agent has of what she is doing—something much less demanding in terms of psychological resources. In this elementary and undemanding sense, intention is not always a matter of what the agent explicitly considers, but mostly of the set of background assumptions that she takes for granted while acting, even if she never explicitly gives them a thought. ICA does not imply that achievements have to be planned in advance, be the effect of any distal intention (Mele 1992, p. 144), or the result of any independent intellectual act of telling oneself what one is intending to do. ICA only points to the idea of an aim, which is what the agent would consider as what she is intending to do while acting if she were asked to explain it, even if this awareness did not take place as some psychological occurring state during action, and only showed up after conscious deliberationFootnote 9. An agent’s performance being intentional in this extremely basic sense implies that she is aware of doing something, an awareness that takes place against the background of certain presuppositions. Intentional action, in this sense, seems to be a much simpler concept than the one of voluntary action, if we understand the latter as the result of some sort of conscious deliberation based on the consideration of reasonsFootnote 10.

In order to make sense of ICA, the notion of intention that we need may be almost as simple as the one involved in our understanding of the purposeful behaviour of relatively basic animals: something along the lines of what Harry Frankfurt (1978) points out when he claims that even a spider could be considered as ‘purposefully’ moving its legs in normal conditions, in the basic sense that its movements respond to some goals, they are controlled by the organism itself, and those attributions of intention may only make sense under certain descriptions of what it is doing. Just as we do not require the spider to be aware of itself in any strongly intellectual sense in order to make sense of its actions in purposeful terms, we would not over-intellectualize the cognitive performances of agents by claiming that they may only be considered as intentional insofar as they respond to the agent’s own goalsFootnote 11.

We would presumably require much more than this to account for fully fledged cognitive performances: something like sensitivity to norms and awareness of the normative features that constitute the context of evaluation where the agent is performing (Broncano and Vega 2011). But even if all of this had to be added to an agential account of epistemology at some later point, all I will need for my present argument is a much more basic requirement for achievements in general, and cognitive achievements in particular: that the alleged success must constitute a goal pursued by the agent. I don’t think we can make sense of the notion of achievement without this basic requirement, and further normative sophistications would have to preserve this idea in one way or another.

3 Back to epistemology: intention in cognitive performances

In this section I will apply those claims to epistemic issues, but let me first consider a case of intentional action that is not related to knowledge:

COLLECTION: Cecile is worried about the poor condition of her nephew’s school, impoverished after years of severe cutbacks. One day, her nephew asks her to give some money for a collection that is being taken up, whose aim is to improve the school facilities. She happily contributes to the collection, unaware that its particular aim is to build a new gym. However, had she known that the money was being collected for this particular cause, she would have refused to participate in it, considering it a misguided decision. She thinks that the gym is in a perfectly acceptable condition, and that money should be spent on many other matters instead, which she considers more urgent, such as improving the state of the classrooms, building up a decent library, or acquiring school supplies.

If we had to decide whether Cecile’s contribution counts as her achievement or not, we would have to split hairs. Defined in a coarse-grained way, her attempt

  1. (a)

    to help the school

was something she succeeded in, and this success was an effect of her performance in no wayward or lucky way. Furthermore, it is an exercise of her moral virtues, generosity in particular, and an effect of her monetary capacities. For these reasons, there is some sense in which Cecile achieved (a). However, as a matter of fact, that same performance may be described in more fine-grained ways, and those descriptions may respond to different intentions. For instance, Cecile could have tried:

  1. (a1)

    to help the school to build a new gym.

  2. (a2)

    to help the school in any other way but (a1).

If we pay due attention to the set of presuppositions she is acting on, we realize that the correct fine-grained description of the action Cecile intends to perform is (a2), and not (a1), even if she did not explicitly consider her goal under any of those two descriptions. The fact that she would have refrained from acting, had she realized that it was (a1) what she was doing, shows that she is presupposing her action to be something along the lines of (a2). Unfortunately for her, (a2) is not something she is actually doing at all, simply because it is not the case that she is helping the school in any other way but to build a new gym. On the other hand, even if she may be described as ‘succeeding’ in (a1), that is not something she achieved, given that she is not intending (a1), which is then a description of her performance that fails to meet ICA.

Cecile is thus ‘succeeding’ in (a1), but not intending to do it; and she is intending (a2), but not succeeding in it. In consequence, none of those two fine-grained descriptions may count as her achievement. What seems like an achievement from the coarse-grained description (a) shows up as no achievement at all from either one of the two proposed fine-grained ones. Nevertheless, it would be mistaken to conclude that this prevents (a) from being Cecile’s achievement: on the contrary, the moral of the story is that, as I said earlier, attributions of intentions to actions fail to meet extensionality tests, and are thus restricted to intensional (with an s) contexts.

Getting finally back to epistemology, what I would like to do now is to take a closer look at cases of environmental luck, such as BARN, in the light of cases such as COLLECTION, which I find relevantly similar. Consider Barney’s cognitive performance: his alleged success could only be described in a very loose way, as something like:

  1. (b)

    to determine whether there is a barn in front of him.

But such a coarse-grained description defines a set of performances that could be taking place in many different contexts. Let me for simplicity split them into two exhaustive categories: ordinary counties and fake-barn counties. Either Barney is performing in a county full of fake barn façades, or he is performing in a county where most objects that look like barns are real barns. In this way, Barney’s performance could be described either as an attempt

  1. (b1)

    to determine whether there is a barn in front of him in a fake barn county

or

  1. (b2)

    to determine whether there is a barn in front of him in an ordinary county.

We would probably choose (b2) as the description of what Barney intends to do since, by stipulation, he does not know, or even suspect, that he is in a fake-barn county. As a matter of fact, the fine-grained description under which his performance could count as ‘successful’ is (b1), given that he is in a county full of fake barns; but this fact is unnoticed by him, and he would be doubtlessly surprised if he later realized that he had ‘succeeded’ in (b1), considering it as a lucky guess, and not as a description of what he was intending to do. Just as Cecile would have refrained from contributing to the collection, had she been aware of its real goal, Barney would have refrained from forming the belief that there was a barn in the field, had he known that he was in a county full of fake-barn façades. In fact, if he had not refrained from forming his belief while being aware of those inhospitable conditions, we would have rightly accused him of being epistemically reckless. And he is not epistemically reckless by stipulation, because otherwise we would not have considered him as rightly justified while forming his belief.

Therefore, we seem to have an achievement from the coarse-grained perspective (since Barney both intends and succeeds in (b), at least apparently), while nothing seems to be achieved in either of the two fine-grained descriptions of that same action: either Barney ‘succeeds’ in (b1), which is not what he intends to do, or he fails in (b2), which is what he was intending to do. Cecile and Barney are thus in very similar situations: only from the coarse-grained perspectives may their actions be considered as achievements, but those achievements are limited by the scope of their intentional action, which is itself limited by the intensional scope of their respective awareness of what they are intending to do.

What this argument attempts to show is that the question whether Barney’s performance still represents an achievement is ambiguous. In order to give it an answer, we have to identify first what description of the action is the one under which its results may count as achievements, which itself determines the proposition we are claiming Barney to know. The extent of his knowledge would be limited by the intensional (with an s) context of Barney’s intentional (with a t) action.

If we combine this conclusion with strong virtue epistemology, we find out that Barney’s knowledge may only be extended to those perspectives under which his cognitive performance still represents an achievementFootnote 12. Imagine the following three JTBs as candidates to describe Barney’s putative knowledge:

  1. (b’)

    There is a barn in this field.

  2. (b1’)

    There is a barn in this field, which is in a fake barn county.

  3. (b2’)

    There is a barn in this field, which is in an ordinary county.

Considered as the results of cognitive performances, (b’), (b1’) and (b2’) have different etiologies, since they may be explained by appealing to different intentions—(b), (b1) and (b2) respectively. By combining my previous description of Barney’s performance in intentional terms and the main tenet of strong virtue epistemology, we obtain the result that Barney knows proposition (b’), while he does not know (b1’) or (b2’). He does not know (b1’) because he did not achieve (b1), given that (b1) fails to meet ICA. And he does not know (b2’) because (b2’) is simply false: although there is a barn in the field, it is not in an ordinary county. The only candidate for knowledge then is the coarse-grained JTB (b’), which is achieved by Barney as a result of intentionally succeeding in action (b).

Now then, is Barney’s JTB (b’) actually knowledge? My position in this respect is that it is. Unlike (b1’) or (b2’), (b’) is a genuine case of knowledge insofar as it is the result of action (b), which is a description under which Barney’s action is both successful and intended, the result of his abilities, and thus his achievement. But how is this possible? Am I not making quite a counter-intuitive claim here? I would say that I am not, and the reason for this may be found in what is known as the generality problem (i.e., the fact that we have to determine the set of contexts where the agent may be considered as reliable)Footnote 13. Here is my rationale for this: it is hard to deny that there is a presupposition, implicitly assumed in BARN, which is that fake-barn counties are something rare and unusual: the sort of place where Barney may not easily expect to find himself. This means that, taking the whole world as the scope of his performance, he is still perfectly reliable in forming the belief (b’), since there are not many fake barn counties in the world (perhaps only one), and thus not many barn-looking objects that are not real barns. For this reason, it seems sensible to claim that his coarse-grained belief may count as knowledge. He knows (b’) because he achieves (b), which is an action he both intends to perform and succeeds in performing—and furthermore he is quite reliable in it. He is reliable if we consider that he is in no particular place, and it is under this description that he intends to perform (b)Footnote 14.

Summarized, here are the main claims that would allow me to deal with cases of environmental luck:

  1. (1)

    Only intentional actions may be considered as achievements (ICA).

  2. (2)

    If some JTB is a cognitive achievement of the agent, then it is knowledge (strong virtue epistemology).

  3. (3)

    There is a coarse-grained description of Barney’s performance under which it is a cognitive achievement, and therefore the resulting JTB is knowledge.

  4. (4)

    Neither of the proposed fine-grained descriptions of Barney’s cognitive performance does represent achievements, since one of them is unintended, while the other one is unsuccessful. None of those does yield knowledge hence (in particular, Barney does not know that “there is a barn in this field, which is in a fake barn county”).

  5. (5)

    Barney’s coarse-grained JTB being knowledge is compatible with his failure to achieve fine-grained knowledge. The result is that Barney knows there is a barn in the field, although he does not know there is a barn in the field of a fake barn county. This is because his belief is an effect of his intentional action under the former, but not under the latter description.Footnote 15

4 What would robust virtue epistemologists earn with ICA?

As I said in section one, my intention is to deffend the core claim of robust virtue epistemology, while remaining neutral on the internal disputes among its different versions. In particular, I believe that my proposal is compatible with the two varieties I have been considering here, since it adds to them a new perspective that, to the best of my knowledge, they had been lacking until now. In this section I will try to make this point clear by first indicating the way my proposal fits respectively into Sosa’s and Greco’s robust virtue epistemologies, and then by considering of some other responses to Pritchard’s argument that are already in the literature.

4.1 Sosa and the risk of infinite regress

According to Sosa (2007, pp. 92–112), Barney would only have animal knowledge, because he ignores some facts about his own reliability in the context where he is performingFootnote 16. Had he known these facts, his JTB would have been upgraded to the much more valuable level of reflective knowledge. What Barney lacks then, according to Sosa, is some factual knowledge about himself and about the world, pieces of knowledge which may be obtained in different ways (perception, memory, testimony, and so on), and would justify an attribution of meta-knowledge (the agent would know reflexively that he knows animally). In later developments, Sosa (2011, p. 92) has introduced a third level, which would require not only the belief to be aptly produced (animal knowledge), or it being meta-apt (reflective knowledge), but furthermore it ought to be apt as a result of being meta-apt (full knowledge). The agent’s getting it right would manifest her competence, she would have aplty considered her own competence in that situation, and her getting it right would manifest such a consideration.

Sosa’s view is open to the objection that these requirements iniciate an infinite regress. If what Barney lacks in order to attain reflective, or even full knowledge, is more factual knowledge about the conditions where he is acting, one could then ask whether such knowledge ought to be merely animal, reflective, or even full. To solve this objection, my proposal is that what he lacks in order to know full well is not just some further factual knowledge, but practical knowledge, i.e. the sort of knowledge that the agent has, qua agent, on what he himself is doing. This allows for a new justification of its origins, and a new rationale on the way it fits into the scene. The reason why Barney’s epistemological deficits are important to evaluating his knowledge is that his action may not count as intentional under the fine-grained description (b1)—i.e., he is not intending to act under those inhospitable conditions and, more precisely, he already knows, at least implicitly, that that is not what he is trying to do. Introducing, as Sosa does, some further knowledge (about the world and about the agent’s reliability in those conditions) as a requirement for the higher degrees of knowledge, seems to initiate an infinite regress; but, in contrast to this, my proposal is that the key to understanding cases of environmental luck is in the agent’s practical knowledge, which is knowledge about what she is intending to do while she acts. In this way, Sosa’s account could appeal indirectly to the agent’s knowledge about her environment, or about her own reliability in it, as a feature resulting from the agent’s practical knowledge, i.e., her awareness of what she is trying to do.

To make it clear: I do not deny that the sort of factual knowledge Sosa requires for reflective knowledge may be crucially important. What I claim is that there is a reason for it, namely that that sort of factual knowledge (about one’s own reliability in a given context) allows the agent to form the right intentions in a responsible way, by allowing her to identify the scenario where she is acting. In my opinion, the key for evaluating Barney’s performance as reflective is not directly in his JTBs about his own reliability in a given context, but in the question of whether he intentionally performed under a description that includes those inhospitable conditions, or he did not. What gives us a basis to evaluate Barney’s epistemic responsibility or recklessness is the scope of what he intends to do in his cognitive performance, i.e. the intensional scope of this practical knowledge, at least up to the point that it affects the safety of his performance. Assuming this, Sosa could combine ICA with the anscombean claim that practical knowledge is of a non-inferential kind (Setiya 2008), or attained without observation (which is a thesis I would endorse, but did not employ or defend here), in order to dismiss accusations of circularity and infallibilism. He could thus prove that he is not requiring knowledge in order to attain knowledge, beginning a never-ending vicious circle. The basis to upgrade animal knowledge to higher degrees of knowledge would not require further evidence or support, since it would be immediate, non-inferential, non-observational, practical knowledge—i.e., knowledge of what one is cognitively doingFootnote 17.

4.2 Greco on fine grained abilities and practical interests

With respect to Greco, I think that my argument would allow him to reject the sort of objection Pritchard (2008) made against him. Remember Greco’s (2007) attempt to solve environmental luck cases by making use of fine-grained definitions of abilities. It was, as I said earlier, the object of Pritchard’s criticism (2012, p. 268), who considered those fine-grained definitions as an artificial and ad hoc movement that would force us to make some controversial revisions of the folk concept of ability. Pritchard seems to be right in that we do not normally define abilities in such a fine-grained way as to indicate very specific conditions of performance—such as ‘having the ability to identify barns in counties that are full of fake-barn façades’. Such detailed descriptions are alien to our everyday practices, and one may rightfully doubt whether, once they are so redefined, we are actually employing the very same concept or not. This would imply that perhaps our new artificial invention lacks the benefits of the original concept, and thus its potential explanatory power.

As I said in a previous section, I myself find Greco’s argument quite unobjectionable, since I consider that attaining sensible revisions of our folk concepts might precisely be one of the main tasks of philosophy as a discipline. But be that as it may, the argument based on intention that I propose here does not rely on any fine-grained definition of abilities à la Greco, but on fine-grained descriptions of the actions performed, a step that seems to be perfectly sensible even from the folk perspective, since it is nothing more than a spontaneous way of describing what the agent may be doing. Considering the scope of intentional action is something we systematically do in order to evaluate the agents’ merits and responsibilities in the results of their actions. In everyday life, an agent’s lack of awareness of a specific description of her action could easily affect the way she may be held responsible for what happened, and praised or credited for her putative achievements. We would not hold Cecile responsible for contributing to building a new gym if she presupposed that the collection had a different aim, and we would not consider Barney as epistemically reckless in forming his JTB, if he were unaware of his epistemically inhospitable environment.

ICA cannot therefore be accused of being some artificial stipulation on the nature of abilities as a natural kind, since it is squarely in line with the way we make credit attributions for success in ordinary life. Perhaps abilities cannot be defined in Greco’s fine-grained way without being quite revisionary; even if this were so (which I don’t think it is), the actions themselves may be described in such a fine-grained way without being committed to any particularly revisionary project. Claiming that Barney has the ability ‘to identify barns in a fake barn county’ is one thing; describing his performance as that of ‘identifying a barn in a fake barn county’ is quite another. Although the former may be considered as some spurious and theoretically laden definition of the ability, the latter is nothing more than a naïve and harmless description of what Barney may be intending to do. Those descriptions of his action that he does not believe he is doing, or he would sincerely reject, after some reflection, as what he was trying to do, should not count as performances he is doing intentionally. And my claim then is that those descriptions may not count as his achievements either. Once supported by ICA, the fine-grained descriptions of the actions would fit quite naturally into our understanding of the scene in lay terms. And it would also fit into the scene as a psychologically plausible requirement, insofar as the Barney’s performance is guided by his awareness of what he is doing, which is a constitutive part of his rational action.

Besides, Greco’s views may benefit from ICA since it helps us realise that it is the scope of the agent’s practical knowledge (her knowledge of what she is doing, or at least intending to do) that allows us to evaluate her abilities in one set of contexts or another, giving us thus an answer to the generality problem. Greco (2007, p. 62) tries to solve that problem by appealing to practical interests, and Greco (2012, p. 23) invites us to evaluate attributions of knowledge keeping always in mind the “relevant practical task or broader practical environment”, although it is usually under-described in the cases considered by epistemologists. Those are moves I am absolutely sympathetic with but, nevertheless, I think we will not understand the role practical interests play in the scene, and the reasons why they affect our appraisals of cognitive performances, unless we consider the agent’s actual intentions, and her awareness qua agent of the scene and the task she is confronting. It could be the case that she is not acting in her own practical interests, or is unaware of them; or it could be the case that she is not paying due attention to the features of the environment that accomplish an important role in the practical task that she is alleguedly performing. In cases like this, it is not practical interests that determine the scope of the agent’s reliability, but the limits of her practical knowledge. In other words: if practical interests have something to say in our deliberation on knowledge attributions, it is because they play a crucial role in the agent’s practical reasoning.

4.3 Littlejohn on the compatibility between luck and ability

I would like to discuss now some deffences of robust virtue epistemology that may be found in the literature already. I am sympathetic to those approaches in that they focus on the definition of “ability” and aim to determine what may count as its genuine manifestation. For that reason, I think my proposal ought to be understood as a complement to theirs, which would be incomplete unless they paid due attention to the role of intention in a normative conception of epistemology.

Let me first focus on Littlejohn (2014)—although a similar argument could be applied mutatis mutandi to Jarvis (2013). Littlejohn puts forward a provocative distinction that crosscuts the divide between robust and modest virtue epistemologists. In his view, the crucial debate is between those authors—like Pritchard or Sosa—that consider the exercise of abilities as compatible with environmental luck, and those—like Greco, Littlejohn himself or, I may add, Jarvis and Carter—that deny such compatibility. This is an inspired way to unblock the debate, and I think that Littejohn faces that task in the right way—that is, by considering the exercise of abilities as something that must take place under the right conditions of opportunity. However, in so far as Littlejohn still seems to assume that the manifestations of abilities are events (and not just descriptions of events), he neglects the importance of the intensional scope of action. If I am right, any achievement (that is, any successful manifestation of an ability) is constrained to those descriptions of the event under which it may still be considered as intentional. Not giving due attention to this limitation produces a sort of ambiguity that explains the different intuitions that lead to the compatibilist/incompatibilists divide. Once freed from that ambiguity, we may find that those on both sides of the divide are getting something right: compatibilists would be right in that there is some description under which the performance may count as an achievement (as it belongs to a wide range of contexts where the agent is reliable), while incompatibilists would be right in that there is another description under which it may not (since it also belongs to another narrower range of contexts where she is not reliable). In order to decide which of those contexts is the one to be considered (that is, in order to face the generality problem) we have to take the agent’s awareness of what she is doing into consideration. It is not enough to consider the opportunity in abstract, as a feature of the event, but the opportunity as a description of the performance under which such event could be considered as intentional (that is, the opportunity as the situation where the agent, qua agent, is intending to act). Since Littlejohn does not do this, the divide he proposes runs the risk of being dangerously ambiguous, and his account still lacks something that is needed to deffend robust virtue epistemology from Pritchard’s attacks.

4.4 Carter on agent-focussed versus belief-focussed success

I will finally discuss J. Adam Carter’s contributions (2011; 2014), which are similar to mine in that he invites us to consider the performance of an agent from different perspectives in order to evaluate whether she deserves credit for it, or her success was just a matter of luck. Carter’s approach is only based on a modal account of luck, and does not resort to the notions of intentional action and practical knowledge, which are the main suppports of my defence. To begin with, Carter (2011) is an attempt to account for environmental luck cases with the resources of robust virtue epistemology by considering a distinction between agent-focused success and belief-focused success. The former would be the success that S—say, Herny—formed a certain belief, which is a matter of mere situational luck (perfectly compatible with knowledge). The latter would be the success that the belief (that S formed) is actually true and, if this succes were lucky, it would be a case of veritic luck, which is knowledge-undermining. Cases of environmental luck would not be lucky from an agent-focused point of view (which would explain why they are still creditable to the agent’s abilities) but from a belief-focussed one (which would explain why they fall short of knowledge). In other words: the fact that that Barney formed such a belief was not lucky, but creditable to his abilities; while the fact that Barney’s belief was true was purely lucky, and this would preclude it from being knowledge. If Carter (2011) is right, belief-focused lucky success would be precluded by an ability thesis, and we could thus account for cases of environmental luck by means of robust virtue epistemology. Pritchard’s additional anti-luck clause would be pointless—as would be, one may add, ICA.

While Carter (2011) shows that robust virtue epistemology could be luck-proof, Carter (2014) faces the task to prove that it actually is luck-proof. And he does so by enriching his preceeding approach with the idea that a belief being creditable to the agent is not just a matter of it being a manifestation of his abilities instead of a manifestation of luck. As Carter correctly deffends there, it is much more sensible to consider this as a gradient relationship, since the intervention of luck is always a matter of degree. In this way, Carter claims, the attribution of knowledge should be done by estimating whether the correctness of S’s belief that p depends (sufficiently) on S’s cognitive ability—in the sense that it depends on S’s ability more so than luck that S’s belief that p is true. That being the case, the truth of the belief would be a modally robust feature, sufficiently dued to the agent’s abilities. In this way, Carter (2011) and Carter (2014) could be nicely combined in order to show that robust virtue epistemology already is environmental luck-proof.

In my view, Carter’s arguments are perfectly right but, just like Littlejohn’s, they still lack from something important in order to account for these cases in a non-ambiguous way. The problem again is that Carter’s modal evaluation of the putative success also takes the event as object of evaluation, and not an intensionally limited description of that event—i.e., the one the agent would consider as what she was intending to do. This is a defficiency that dooms Carter’s estimations to ambiguity. It is not enough to consider the event by focussing on one of its aspects (either the situation where the agent acts, or the truth of her belief): we still have to say which description of the event is the one we are considering, and such a decission would affect both agent- and belief-focussed estimations of luck in a dramatic way. Remember that we had to evaluate Barney’s achievement as a result of an intention, and we had three alternatives there: (b), (b1) or (b2). Choosing each of those alternative descriptions should be previous to the decission to focus on one aspect of it or another. If we choose (b), for instance, we can then either focus on the agent or the belief, and weight the relative effect of ability and luck, or we could focus on its truth, by estimating the modal robustness of (b’). And the same alternatives would appear if we choosed (b1), or (b2). We should first take a description of the event, and then decide whether to focus on the agent or the belief, and our first choice is crucial because it completely transforms the set of possible worlds in contrast to which the performance is going to be modally considered. Unless we constrain our evaluations to the intentional description of the action, we could be crediting Barney (or denying him credit) for his success in something he is not intending to do at all. I would thus say that Carter’s robust virtue epistemology is on the right track to be anti-luck, but it would have to put its resources to work on the estimation of intensional descriptions of events, and not on events as suchFootnote 18.

5 Some objections

Most of the objections to my argument that I have encountered are related to ICA, and attempt to undermine the idea that only intentional performances may count as achievements. Although ICA is not a claim about epistemology in particular, but about the theory of action, it certainly is the most crucial—and also perhaps the most controversial—point on which my whole argument hinges, and I am aware that it may be disputed in a number of ways. In the rest of the paper I will consider some of them, and try to overcome them.

5.1 Putative achievements that are not intentional actions

At first glance, there seem to be many events that happen to us, which may be described as achievementsFootnote 19. However, these are not properly things that we do, and thus cannot be considered as our intentional actions. Think for instance about putative achievements such as being elected president of a country, being taken on by a company, earning a PhD, or even having a happy life. Although those are not performances of the agents, we do plausibly consider them as achievements. If this intuition were right, the attack on robust virtue epistemology could be reconstructed by claiming that this is the sort of achievement that yields knowledge: successful occurrences that are not intentional actions.

Let me focus first on the case of a PhD: although earning a PhD seems to be an achievement, it would be odd to claim that it is something one does intentionally: it is awarded to the person, not the effect of any immediate performance on her part. My response to this sort of objections would be twofold, since we should distinguish between ‘earned’ degrees and honorary ones. The former are the effect of many intentional performances of the candidate: her decision to earn a PhD, her application to the program, the design and implementation of her research project, the public defence of her dissertation, and so on. I would say that, in the end, the degree is ‘earned’ by the agent as an effect of this complex set of temporally structured and planned actions, which do in general respond to the pattern of intentionality. In fact, such complex and temporally structured planning may be the key to understanding the concept of intention in the light of the ones of autonomy and rationality (Bratman 1987). Earning a PhD is not an intentional action itself, but it seems to be essentially constituted by many actions that have to be intentional, making up a coherent and stable plan that is pursued by the agent along time. Even if we considered the PhD itself as an achievement beyond those other actions, intention would still be a necessary component of the scene.

The rationale would have to be slightly different in cases of honorary degrees or awards, since candidates normally do not propose themselves for them, nor do they normally plan and structure their actions in order to attain them. Perhaps the candidates never envisaged the possibility of being awarded such a prize, never performed any action with the explicit or implicit intention of obtaining it, and would be extremely surprised when finding out about it (just as Cecile or Barney would be surprised to know that their actions could fall under those non-intentional descriptions). In this case, my response would be that honorary degrees are a way of recognizing the success of a person in some relevant performances that were themselves intentional: deeds that led to some significant advance in a field of research, or an important contribution to some social good. While it is correct that honorary degrees are not awarded in response to intentional candidacies, they represent recognition of the achievements obtained by the candidate. Insofar as it is not intentionally pursued, I would hesitate to consider the degree itself as a further achievement: I would see it rather as a way to praise previous achievements of the agent, which does not have to represent a further achievement itself.

Similar bifurcated rationales could be proposed for other alleged achievements that are not intentional actions, such as becoming president of a country or being taken on by a company. But perhaps this line of argument could not be so easily applied to the case of having a happy life. Consider:

HAPPY LIFE: Joy is a woman that is praised for attaining happiness in life even if she merely focused on small tasks and never considered or entertained the broad philosophical question of whether and how she should attain a happy life (on the whole)Footnote 20.

Could that happy life be considered as Joy’s achievement? My view is that hardly would we do so if we did not consider attaining a happy life as one of the most general aims of human beings in general, and thus of Joy in particular. And that could be in fact Joy’s aim even if she never reflected about this. Remember that the notion of intention employed in ICA does not require any explicit act of forming the intention to take place in the head of the agent. No psychological process of representing what one is about to do ever has to occur. On the contrary, all that is needed is an implicit presupposition of the scenario one is in, and the kind of task one is performing. And perhaps those assumptions only become explicit (if they ever do) after conscious deliberation, once certain possibilities are considered by the agent. If, reflecting on the ends of life, Joy sincerely denied that attaining happiness ought to be one of our goals—but, say, she claimed that we ought to live in miserable pain in order to pay the penalty for our sins—, hardly would we claim that what she attained in HAPPY LIFE was something she achieved.

5.2 Extremely precise descriptions of achievements

There are very common situations where we qualify as ‘achievements’ descriptions of actions that are not explicitly pursued by the agents, or considered by them as what they were exactly intending to do. Here is an example:

FAST RUN: Usain Bolt achieved running 100 m in 9.58 s in Berlin in 2009, breaking the previous record he made himself in Peking in 2008, which was 9.69. Finishing the run in precisely 9.58 was most probably not what he was intending to do in Berlin, but it certainly was an achievement of his, and quite an amazing one.

Could we not make the same claim with respect to Barney? Could we not consider his identification of a barn in a fake barn county as a precise description of his achievement, beyond what he was consciously entertaining as a possible result of his own action?

With respect to FAST RUN, I would say that finishing the run in exactly 9.58 is an achievement because it is an extensional description of an action that was intentional under another description, namely: to finish a 100 meter run faster than anybody else had ever done. That precise description of Bolt’s performance was not pursued by Bolt as such, but it was his achievement because it is an exact extensional description of what he actually attained. His performance was intentional under the coarse-grained description of “finishing the run in less time than anybody else in history”, and it was not intentional under the fine-grained description of “finishing the run in exactly 9.58”. However, “finishing the run in exactly 9.58” is a description of Bolt’s achievement because it is an extensional description of the actual result of his intentional performance.

In fact, if we took the precise description of the performance at face value, we would find out that, insofar as it was not exactly intended as such by the agent, it may not count as his achievement either. Compare FAST RUN to the following situation:

PRECISE RUN: Bolt had claimed to be able to run 100 m in any previously given time within a certain range with a precision of 0.01 s. In particular, he had claimed that he was able to finish the run in exactly 9.58—and not in 9.57 or 9.59. He actually did, and even proved that he was quite relibable in doing this.

In PRECISE RUN we could say that Bolt ran 100 meters in 9.58 intentionally, and that would be a still more incredible achievement than the one attained in FAST RUN. But this would represent quite a different achievement. Bolt’s achievement in FAST RUN was to finish 100 meters in less time than anybody else had ever done, i.e. 9.58 seconds. But, in that case, finishing in 9.58 seconds was not an achivement at all with respect to the task of finishing a 100 meters run in a given time with a precision of 0.01 seconds.

Was “finishing the run in 9.58” Bolt’s achievement or not? The moral of these examples is that we may find no answer to this question if we only look at a watch: we have to take the agent’s goals into consideration, which are themselves defined by the task he is intending to perform. Take any description of the agent’s performance you may consider as his success: the question as to whether it is a genuine achievement or not cannot be answered unless we consider the way that result is related to what the agent intended to do, the scenario he was recognizing for his own performance, and the task he considered he was undertaking. It does not matter whether we extensionally identify the result in a way that is infinitely more precise than the agent’s awareness: our assessment of that extensionally precise description as a genuine achievement may not go beyond the agent’s awareness of what he was doing or, at least, trying to do.

5.3 Putative achievements not foreseen by the agent

Another possible objection to ICA could claim that, although there is some conceptual connection between intention and achievement, I have been opting for a too strong claim, one of strict identity, while perhaps we should weaken ICA in order to make it more acceptableFootnote 21. Remember that ICA was:

ICA: Only intentional actions and their intended effects may count as achievements.

Now, the weakened alternative could go like this:

ICA*: Only events that are intentional actions under some description, and states or objects that are the effect of intentional actions under some description, may count as achievements.

ICA* is not as strong as ICA, since we could still consider as an achievement a description of an event that is not itself intentional insofar as there is another description under which it is intentional. Or we could consider an object as an achievement even if it was not the intended result of an action, but merely an unexpected effect that the agent attained while pursuing some other goal.

Opting for a weaker thesis like ICA* could be a way to defend Pritchard’s claim that agents performing in scenarios affected by environmental luck may attain achievements, even if those achievements were unexpected by them. For instance, Barney’s successful recognition of a barn in the fake barn county would still be an achievement of his, even if he is unaware of being in this situation, because his cognitive performance would be intentional under another description (i.e., he would have the intention of identifying a barn tout court). ICA* would preserve the core of Pritchard’s argument against robust virtue epistemology, accommodating an important part of the intuition that supports ICA.

Nevertheless, in my opinion, ICA* is too weak. In order to show this, I am going to focus on two further examples employed by Pritchard himself, which allegedly represent achievements in contexts affected by environmental luck. Consider first:

PIANIST: Carla is playing the piano in a room that, unbeknownst to her, is surrounded by water. The walls could completely give way and let the water in at any moment. By pure chance the walls hold up, and Carla plays masterly.

According to Pritchard, intuitively, the ability that Carla would be displaying in this situation is exactly the very same ability she would display while playing the piano in normal environments. This counter-argument of Pritchard is worrisome for my proposal, because agents in COLLECTION or BARN are unaware of some aspect of their respective environments, and this feature is what allows me to deny that their actions could be considered as achievements from those perspectives. Pritchard could claim that, just as Carla’s performance still represents an achievement, despite its lack of safety, Barney’s performance still constitutes an achievement despite its lack of safety. Those agents would not intend to act under those conditions, but they would have still manifested the very same abilities they would have shown in normal conditions.

I would reply to this objection by giving the same medicine to Carla that I gave to Cecile or Barney. Her intention may be considered under a coarse-grained description, such as

  1. (c)

    to play the piano

or under either one of those fine-grained ones:

  1. (c1)

    to play the piano surrounded by water that may come in at any moment.

  2. (c2)

    to play the piano in ordinary conditions.

I think that Pritchard is being ambiguous when he claims that Carla is stillmanifesting the very same ability she would be manifesting if she were playing under normal conditions. The fact is that (c1) and (c2) are very different activities, which might require very different abilities, as Greco defends. However, I think that Greco’s position in this respect is open to Pritchard’s criticism unless Greco assumes ICA, because we would still not have the relevant information on PIANIST, if we didn’t take into consideration the agent’s awareness of what she is doing.

The crucial question for making sense of PIANIST is whether Carla would still be reliable in her performance under some given description, if she were doing it intentionally under that same description. To test this idea, let me consider the following tweak for PIANIST:

SHY PIANIST: Although she is an excellent pianist, Carla is extremely shy, and gets very nervous whenever she plays in public. In order to allow her overcome this shyness, her piano teacher Wolfgang invented and implemented the underwater room, which actually produces in her a kind of disturbance that is very similar to the one she would feel playing in front of people. Thanks to Wolfgang’s invention, she managed to control her fears, and to play masterfully under stressful conditions.

For Wolfgang’s invention to work at all, of course, Carla would have to be aware of the conditions under which she is playing and, after some practice, we could imagine her being able to play surrounded by water that may come in at any moment, which, in the end, could perhaps help her acquire the ability to play in front of a public. My point is that what is required to have the relevant ability is not just the reliable capacity to perform under those conditions, which is the way Greco defines abilities. In that way, Greco is exposed to cases such as Pritchard’s PIANIST. What Greco needs here is to assume the implications of ICA. In order to consider the performance as an achievement of the agent, it must be intentional under that same description, which requires the agent to be aware of the situation, and for her to know what it is precisely that she is intending to do, how, when and where she is intending to do it. ICA* would not be sufficient for this, because we would still consider as achievements performances that are described beyond the agent’s awareness. But we would not claim that Carla ‘achieved’ playing surrounded by water if she was not aware of this particular condition of her environment, just as we would not claim that she achieved playing in public if she were only able to do it unaware of the fact that there was somebody there, hearing her performanceFootnote 22. Achievement implies awareness, precisely because it is a requisite for intentional action.

Greco’s appeal to fine-grained conditions of success is insufficient by itself to respond to Pritchard: achievements are not just a matter of being reliable in the performance under finely grained conditions. In order to represent a genuine achievement, the agent must be aware of the conditions under which her action is being considered as successful, which is what allows her to act intentionally under that specific description. ICA* is not enough for this: we require the stronger version, ICA.

The same point may be clarified by considering another example frequently employed by Pritchard (2008:445, 2010a, 35):

FORCE FIELD: Archie is an archer that selects his target from a range of potential targets entirely at random. He skilfully fires his arrow, he hits the target, and his success is not due to any deviant lucky causal chain. Nevertheless, unbeknownst to him, all of the other targets that he could have fired at contained force fields that would usually repel arrows. Fortunately, Archie just happened to fire at the one target that lacked such a force field.

Pritchard introduces FORCE FIELD in order to prove that some performance may count as an achievement, and still be unsafe. Even if Archie could very easily have failed, that does not prevent his performance from being skilful. Pritchard’s conclusion is that, since knowledge requires safety, while ability manifestation does not, an ability condition may not be enough for knowledge.

The problem with FORCE FIELD is, once again, that the putative achievement has not been considered in the light of the agent’s awareness of what he was doing. As a matter of fact, there are many different descriptions of what Archie may be intending to do in the example, such as:

  1. (d)

    to hit a target.

  2. (d1)

    to hit the only target that is free of force fields.

  3. (d2)

    to hit a particular target, previously chosen at random.

  4. (d3)

    to hit the same target that was hit by the previous archer.

  5. (d4)

    to hit the target on row 12, column 6.

    \({\ldots }\)

These and many other descriptions of his action may be considered as what Archie is doing, but not all of them are descriptions he, or anybody else, would recognize as what he was intending to do. My point is that, by the same token, not all of them may be rightly considered as his achievements. If the requirement we had for achievements were just the weak one, ICA*, we would be forced to claim that all of these descriptions represent genuine achievements. However, (d1) is not something Archie would be reliable at, and he succeeds in it by pure chance. It is in fact not something he is aware of doing at all, and had he intended to perform it, we would then have considered his performance as reckless, his success as hazardous, and his alleged achievement as just a bluff, since he had no clue whatsoever as to what the right target to choose was.

For that reason, once again, I think that ICA* is too weak. We need the stronger requirement, ICA, since Archie only achieves what he intends to do: he intends (d), to hit a target, and he succeeds in fulfilling this intention in a non-wayward way—he thus achieves it. He intends (d2), i.e., to hit that particular target, that he had previously chosen at random, and he succeeds in hitting it in a non-wayward way—he thus achieves (d2). But he did not intend (d1), to aim at the only target free of force fields, because he is not even aware of this feature of his environment. His ‘success’ in (d1) may not be explained by his intention to do it. And, for this reason, he does not achieve (d1), which is some alleged ‘success’ of his performance he did not intend to attain. In a similar way, we would have to know what descriptions of Barney’s action are being consciously entertained by him as what he is doing in order to answer the questions whether (d3) or (d4) may represent genuine achievements or not.Footnote 23

Here is a similar potential counterexample that, in my opinion, could be dispelled with the same argumentFootnote 24:

SURPRISE DISCOVERY: Mr. Y is a geneticist: he intends to discover whether gene A has property F. Turns out though, in his investigation, he discovers that gene A actually has a very different, hitherto undiscovered property, Z. Mr. Y’s discovery that gene A has property Z is regarded to be Mr. Y’s ‘greatest scientific achievement’ though it was not Y’s intended aim.

It is hard to deny that there would be a very relevant description of Mr. Y’s performance that ought to be considered as intentional: to discover the properties of gene A. And this discovery could only be rightly praised as his achievement under that description. Contrast this to purely accidental discoveries that were only due to serendipity—like perhaps those of saccharin or LSD. Hardly could we seriously praise those cases as “scientific achievements” of their respective discoverers, if they did not intend to attain those results under any description whatsoever.

5.4 Absorbed agents

Some agents have the ability to abstract themselves from their environment, being able to ignore disturbing conditions surrounding them while performing some difficult task. In these cases, agents might finally not even be aware at all of those conditions while they perform, but we may still rightfully consider their success as an achievement attained under those disturbing conditions. The problem is that, once they fully concentrate on their task, they seem to lack awareness of the fact that they are performing under those conditions, and thus lack practical knowledge on this aspect of their own actionFootnote 25.

I would respond to this insightful counterexample by differentiating this sort of successful performances from the ones attained by agents who are completely unaware of acting under disturbing conditions. For instance, we don’t praise deaf agents for their ability to abstract from environmental noise, just as we saw before, we wouldn’t praise Carla for her ability to play surrounded by water in PIANIST, while she was fully unaware of that aspect of her environment. The situation would be very different for hearing people able to actively ignore environmental noise, or for Carla once she is able to play in Wolfgang’s room disregarding her disturbing conditions. In fact, what Carla gains after practicing in Wolfgang’s room might precisely be the ability to abstract herself from the environment up to the point of not even being aware of it anymore. In this case, I think we would rightfully claim that she achieved playing in that room, even if she was not aware of this while she was doing it.

So I am willing to bite this objection’s bullet, and accept that it is possible for an agent to achieve success in a performance under a certain description, even if at the moment of the performance she is fully unaware of those conditions. Nevertheless, I think this bullet is not lethal for my proposal because we should reconsider the task we are evaluating from a temporally extended perspective. If we paid attention to the case, we would realize that what we are considering as an achievement goes beyond Carla’s actual performance. In my opinion, we claim that Carla achieves playing surrounded by water in SHY PIANIST because we take her action as a unit that extends beyond her actual playing. While playing, she might lose awareness of those conditions, but before starting to play, and probably after doing so, she would be perfectly aware of them—otherwise, her merit would be similar to that of a deaf person allegedly able to abstract herself from noise, and would not be different in the relevant sense from her own situation in PIANIST, when she was simply unaware of her disturbing environment. Her performance under troubling conditions is an achievement of hers because we consider it in continuity with the moment when she decided to abstract from those conditions. Her decision to play in Wolfgang’s room, together with her astonishing act of concentration and the further act of playing piano really well while she is temporally unaware of her situation, constitute a whole and coherent unity, tied together by diachronic plans and policies, and inscribed in a wider narrative. It is that whole which represents an achievement of hers, precisely because her intentional action of “playing in the underwater room” is constituted by this complex diachronic structure.

What this objection shows us is not that there may be genuine achievements deprived of awareness, practical knowledge, and intention, but that those concepts may only make sense within complex, temporally extended structures of agency, plans and policies, as the ones Michael Bratman famously described (1987; 2007).

5.5 Abjurer agents

Some people seem to be unable to recognize their own achievements. Think about a painter abjuring from her own work, as so often happensFootnote 26: that painter might claim that that particular painting is not what she was intending to do, or certain features of it, and she may hence deny that that canvas is a faithful manifestation of her original intention. Perhaps she would even be willing to destroy it, considering it deprived of any value at all, despite the fact that most people would see it as an amazing achievement. Even if the result was not what she had in mind while she was painting it, and therefore it is not what she intended to do, it seems that we could still consider her masterpiece as a genuine achievement. The painter could be wrong, and we, the observers, could be right.

This case is problematic for my argument because, although those actions are intentional, we would consider as achievements some results of those actions that are not themselves recognized as intentional by their agents. If this were possible, my objector could apply the idea to BARN, and claim that Barney achieved (b1)—i.e. finding out that there was a barn in front of him in a fake barn county—, even if he was not aware of having succeeded in this, and would deny that this was something he was intending to do. In other words: the example could show a situation where intention and achievement may come apart, which is what I have been trying to prove cannot happen.

First of all, this case is clearly affected by matters related to subjectivity and taste: what we would consider as valuable according to our standards may be worthless according to the artist’s more demanding standards, or different tastes. Success is always a matter of standards, most times of a normative kind, and those standards go beyond the individual’s evaluation. The agent does not always—or perhaps ever—have the last word judging her own success and, by the same token, her own achievements.

If this is correct, it would be too demanding to require from the agent awareness of her own success in order for her performance to count as an achievement. All that is needed, as I earlier said, is that the agent tries—and knows, in a practical way, that she is trying—to obtain the effect in question, but not that she actually knows whether she is succeeding in it or not. Our artist is trying to succeed in painting a masterpiece, or at least a decent work of art, and were she not doing this intentionally, we would not consider her result as an achievement—although perhaps we could still consider it as art, since the concept of artwork and the one of achievement do not have to be co-extensionalFootnote 27.

To illustrate this, imagine the following scene:

CANVAS: while working on her painting, Anne had left a canvas on the floor with the only intention of preventing the floor from getting dirty. Unnoticed by her, she left a remarkable pattern of drips, footprints, and circular can patches of different colours on the canvas that was on the floor while she was working on her picture.

Somebody could later take Anna’s canvas for artwork, although she would have probably thrown it away as garbage. My point here is that, if we were aware of its etiology, nobody would claim that the canvas on the floor is Anna’s achievement, although we could still attribute to it some aesthetic interest. The canvas could be disturbing, appealing, even fascinating, but not inspired. This shows that our standards of evaluation are—or even should ideally be—focused on features of the object itself, not on its etiology, just like the truth of the belief is a matter of its relation to facts, and does not depend on the way it was attained by the person who formed it. Nora’s belief in SHEEP is not less true because it is lucky, nor is Barney’s one in BARN. For the same reason that lucky true beliefs are not less true because they are lucky, beautiful canvases are not less beautiful because they were unintentionally produced.

What we could claim about the canvas on the floor may also be claimed about features of the painting the agent did not intend to put there, aspects she was unaware that she was expressing, or traits she did not recognize as what she was intending to do. Those unintended features may have contributed to the painting in ways she did not envisage, and perhaps the painting is a much better artwork precisely because of it! However, those unwilled, unexpected and unintended aspects of the painting would not be rightly considered as achievements of the artist, no matter what aesthetic value they may have. This shows, by the way, that there is an extremely complex relationship between the artist’s intention and the results of her work. Perhaps a piece of art is not just something the artist does but at least partially something that happens, even despite the artist’s intention—not to mention the unconscious aspects of her performance, which are not particularly relevant for our case. The value of the artwork is not constrained by the intention of the artist, precisely because it is not constrained by the limits of her achievement. There may be much more value in it than the one that the artist actually achieved.

Getting back to the (apparently) more solid ground of epistemology, this example shows us that, in order to represent some cognitive achievement, it is not necessary that the agent knows that she is succeeding in her performance: it is enough that she knows that she is trying to succeed in it. Perhaps the agent does not know whether she got it right or not, and would have to wait for further confirmation—in any case, that would not affect the fact that her cognitive performance was her achievement. In order to make BARN work as an example of environmental luck, we don’t need Barney to know that he succeeded in recognizing a barn, but just for him to know that he tried to do that. And his success is measured as an achievement by the limits of this trying. Barney was not trying to succeed under inhospitable conditions because he was not aware of them. That description of his performance may not be considered as intentional and, by the same token, that particular ‘success’ was not his achievement, even if it was a remarkable ‘success’ (just as a remarkable piece of art may go beyond the painter’s intention, and thus have more value than her achievement).

6 Conclusion

In a nutshell, my proposal is that only intentional actions and their intended effects may count as achievements, which entails that performances may only be considered as achievements under some descriptions, but not others. Applied to cognitive performances affected by environmental luck, this implies that agents in fake barn cases are not acting intentionally under the description that affects their reliability (since otherwise those cognitive performances would be reckless), and the resulting justified true beliefs are not their achievements under those descriptions.

This would explain why, and to what extent, justified true beliefs attained under unsafe conditions fall short of knowledge. The alleged fact that knowledge and achievements may fall apart in cases of environmental luck would have been proved to be wrong: environmental luck undermines knowledge in exactly the same way it undermines the achievement. If I am right, knowledge and cognitive achievements may not come apart: they hold each other up, or they fall down together, which is the core claim of robust virtue epistemology.