Epistemology Part 3: Sosa and the Speckled Hen

Ernest Sosa, in his book Beyond Internal Foundations to External Virtues, uses the problem of the speckled hen to attack classical foundationalism generally, and internalism about justification more specifically. According to many forms of foundationalism, beliefs are foundationally justified (justified without recourse to any other belief) if they are taken directly from experience. As an example, under this theory, if I look down and see that my keyboard is rectangular I am foundationally justified in believing so. Internalism states that all differences between justification in subjects can be explained with recourse only to current mental states. The problem of the speckled hen states that there are some aspects of experience which cannot reliably be transformed into belief. In the classic case, when one looks at a hen with forty-eight speckles, without counting one is unable to form the justified belief that there are exactly forty-eight. Under my example, if I look down at my keyboard I cannot instantly tell how many keys it contains in the way that I can when ascertaining its shape.  In this essay I will examine three claims made by Sosa against internalist solutions to the speckled hen. The first claim I will be examining is that phenomenal concepts cannot contain arithmetical content. The second claim is that adding the requisite of attention to the classical foundational model is insufficient to solve the problem. The third is Sosa's argument against phenomenal indeterminacy, which I will argue is completely defeated by Michael Tye. Then, I will give an explanation of Sosa's externalist solution to the issue. Finally, I will explicate my own solution to the speckled hen dilemma, which is very similar to one proposed by Michael Pace.

Before tackling Sosa's first claim, I must define the relevant terms. Sosa distinguishes between three different types of concepts: phenomenal, indexical, and simple geometric or arithmetical (SGA) ones. Of the three, only phenomenal and SGA concepts are relevant to this discussion. Phenomenal concepts are those that pertain to phenomenal properties; the aspects of our experience. For example, I am able to recognize my car when looking for it in a parking lot because I have a phenomenal concept of how my car appears to me. SGA concepts are purely informational ones that pertain to mathematical values. For instance, “2+2=4” and “all triangles have three sides” are both SGA concepts. According to Sosa, “SGA concepts differ from phenomenal ones in this respect: no guarantee of reliability in applying them to experience derives simply from understanding them.” (Sosa p.127) This statement is certainly true. Just because I have a concept of the number thirty-six does not mean that I am capable of recognizing thirty-six objects as being such.

With the relevant concepts having now been defined, I will move on to Sosa's hourglass example. Consider an hourglass shaped pattern composed of eleven dots. If you had seen such a pattern before, you would have a phenomenal concept that corresponds to the hourglass. With said concept you would be capable of distinguishing between the hourglass and other similar patterns. According to Sosa, “...only counting can make you justified in believing that there are eleven [dots]. Your phenomenal concept of that eleven-membered array is then not an arithmetical concept, and its logical content will not yield that the dots in it do number eleven.” (Sosa p.128) This is short sighted. If you had counted the number of dots when you first obtained the phenomenal concept, then you could know at a glance that the dots number eleven. Nothing stops the phenomenal concept of the hourglass array from containing the information that it has eleven dots in the same way that nothing stops my phenomenal concept of my car from containing the information that it has four wheels. The claim that phenomenal concepts cannot contain arithmetical data is not important for Sosa's solution to the problem, however its rejection is extremely important for my forthcoming solution to the speckled hen issue.

Sosa's claim discussed in the previous paragraph is part of his argument against what he dubs modified classical foundationalism. He describes this modification as such, “An SGA belief that one's experience has feature F (an SGA feature) is foundationally justified so long as (a) one's experience does have feature F, (b) one believes that that one's experience has feature F, and (c) one attends to feature F.” (Sosa p.129) Although I did not agree with the previous claim, I do agree with Sosa that this system is insufficient for solving the speckled hen problem. To properly debunk the theory I will be modifying Sosa's decagon example (p.129) slightly. Imagine  that you are seated in a room and are clearly shown a white decagon against a black background. Also imagine that you have no phenomenal concept of “decagon,” such that you would be unable to distinguish one from other similar shapes. You may have an arithmetical concept of “decagon,” but as you are not given the opportunity to count the sides of the figure, this is unimportant. Finally, imagine that the words “this is a decagon” are displayed above the shape. Under these circumstances it is quite possible for all three of the previously stated requirements to be met, and yet you are not justified in your belief that you are viewing a decagon. Your experience contains “decagon-ness,” you believe that you are viewing a decagon, and you are attending to said decagon. Yet, if the shape presented was instead a dodecagon, there would be no corresponding change in your belief.

In Sosa's lead-up to his own solution to the speckled hen, he next turns on the solution involving phenomenal indeterminacy. Phenomenal indeterminacy is the idea that under certain circumstances, our experience doesn't represent a specific (or determinate) feature. In his argument against this idea, Sosa turns away from phenomenal concepts and towards occurent thought. There are two occurent thoughts that Sosa wants us to consider: “(T1) That if squares have more sides than triangles, and pentagons have more than squares, and hexagons have more than pentagons, and heptagons have more than hexagons, then heptagons have more sides than triangles,” (Sosa p.133) and “(T2) That if squares have more sides then triangles, then squares have more sides than triangles.” (Sosa p.133)  Sosa then asks us how many different geometric  figures are contained within T1. Answering this question correctly without counting may not be impossible, but for most people answering without counting carries at least a moderate amount of uncertainty. Despite this there is still a determinate number of figures contained within the thought. How many are there in T2? This question is much easier; there are two. The distinction between these two questions is at the core of Sosa's argument. “Some intrinsic features of our thoughts are attributable to them directly or foundationally, while others are attributable only based on counting or inference. How will the classical foundationalist specify which features belong on which side of that divide?” (Sosa p.134)

Unfortunately for Sosa, Michael Tye has provided a concise and effective answer to this question. The specification is made by the capacity to attend. In other words, beliefs can only be foundationally justified by phenomenal properties that we are capable of attending to. All we need do in order to fit this to Sosa's argument is define a new term. I shall therefore call properties associated with occurent thought introspective properties, and their corresponding concepts introspective concepts. Because we are incapable of directly attending to the number of figures in T1, our introspective experience is indeterminate with respect to the number of figures. In the case of T2, we are perfectly capable of attending to the number of figures, and thus our introspective experience is determinate. In his essay, A New Look at the Speckled Hen, Tye applies this idea directly to the speckled hen case. “... you are conscious of the speckles on the hen without each speckle being such that you are conscious of it. The reason that you cannot enumerate the number of speckles is that the enumeration would require you to attend to each of the speckles [individually]. This you cannot do in a single glance...”

Although I do not think that Sosa successfully defeats the theory of phenomenal indeterminacy, he does make one comment about the theory that I agree with. “If so, it may be claimed, accordingly, that if one lacks a phenomenal concept then one's experience will not in fact have the feature corresponding to that phenomenal concept.” (Sosa p.131) To elaborate, this means that the experiences a person has can actually change according to the phenomenal concepts possessed by that person. This is a result that neither Sosa nor I wish to accept, and that is why I do not personally endorse phenomenal indeterminacy.

We have now finally arrived at Sosa's externalist solution to the speckled hen problem, which is actually surprisingly simple. Sosa's aim is to find a reliable way to distinguish between phenomenal experiences that are not capable of providing foundational justification for beliefs, and those that are. Specifically, what distinguishes between the experience of a forty-eight speckled hen (where one cannot directly tell how many spots there are) and that of a three speckled hen (where one can directly tell)? “The relevant distinction is that the latter judgment is both safe and virtuous... It is 'safe' because in the circumstances not easily would one believe what one now does in fact believe, without being right. It is 'virtuous' because one's belief derives from a way of forming beliefs that is an intellectual virtue, one that in our normal situation for forming such beliefs would tend strongly enough to give us beliefs that are safe.” (Sosa p.139) This is obviously not compatible with internalism, as the situation under which we form beliefs is not a mental state. It is well beyond the scope of this essay to explicate all of the reasons to rejects Sosa's solution. Here, I believe that it will be sufficient to say that there is no reason to turn to externalism if a suitable internalist solution to the speckled hen can be found.
    I also believe that such a solution has already been proposed by Michael Pace in his essay Foundationally Justified Perceptual Beliefs and the Problem of the Speckled Hen. This solution is explicated by Pace on page fourteen under the title Recognitional-MFP (Modest Foundationalist Principle). However, the language and terminology used by Pace are unsuitable for my current discussion, so I will be reformulating it in my own words. The modified classical foundationalist principle given by Sosa and mentioned above is still missing two requirement in order to solve the problem of the speckled hen. Therefore, here are my requirements for obtaining foundationally justified belief from experience. A belief that one's experience has phenomenal property F is foundationally justified if (a) one's experience indeed contains property F, (b) one believes that one's experience contains property F, (c) one attends to property F, (d) one possesses a phenomenal concept f that corresponds to property F, and (e) one is capable of recognizing that property F is an example of concept f. As I have already established that phenomenal concepts can contain mathematical data, there is no reason to make reference to SGA concepts. When viewing a hen with only three speckles, we are foundationally justified in the belief that the speckles number three because we have a phenomenal concept of three, and we are capable of recognizing when we are presented with the phenomenal property of three-ness. In the case of the forty-eight speckled hen, we lack both the phenomenal concept of forty-eight, and the ability to recognize forty-eight-ness.

In conclusion, although Ernest Sosa's externalist solution to the problem of the speckled hen works, he by no means effectively proves that there are no sufficient internalist solutions. At the very least, positions advanced by Michael Tye (phenomenal indeterminacy) and Michael Pace (recognitional ability requirements) remain solid despite Sosa's argumentation.