Wednesday, June 2, 2010

Phenomenal Sorites and Conceptualism

This is the 8th and final part of the serialization of the long version of my paper, "Color-Consciousness Conceptualism," the short version of which appeared in the Second Annual Conference of Consciousness Online. This post contains sections 8 and 9 of the paper.

8. Phenomenal Sorites
I turn now to consider whether the conceptualism on offer in the present paper—specifically, the Second Approximation—runs afoul of the hypothesized existence of phenomenal sorites series. Many philosophers have been convinced that, because of such series, indiscriminability is intransitive (Deutsch, 2005; Goodman, 1951; Hellie, 2005; Pelling, 2008). If indiscriminability is indeed intransitive, then this poses a real problem for views such as the First Approximation wherein indiscriminability of two shades is accounted for by the sameness of the color concept applied to each. Sameness of concept applied is clearly transitive and thus cannot be an adequate account of indiscriminablity if indiscriminability is intransitive. That’s the gist, at least, of the alleged problem that phenomenal sorites series pose for conceptualism. Before saying more about the alleged problems and my solutions to them, I first turn to spell out some relevant differences in kinds of phenomenal sorites series.

8.1. The kinds of sorites
A phenomenal sorites series of colors is a set of colors ordered in such a way that each member in a pair of adjacent colors are perceptually indiscriminable, but colors at the beginning and end are perceptually discriminable. One example of such a series would be 34 colors, the first and last of which look unique red and unique yellow, respectively, but each of the 34 colors cannot be perceptually distinguished from its immediate neighbor. The smallest phenomenal sorites series would consist of only three colors, A, B, and C. In such a 3-member series, A is indiscriminable from B, B is indiscriminable from C, but A is discriminable from C.

I will be interested in examining kinds of phenomenal sorites series. The different kinds can be distinguished in terms of two orthogonal dimensions of difference. The first dimension of difference is between diachronic phenomenal sorites series and synchronic phenomenal sorites series. The second dimension of difference is between, on the one hand, series with first and last members conceptualizable as falling under the same noncomparative color determinable, e.g. LIGHT BLUE, and on the other hand, series with first and last members conceptualizable as falling under distinct noncomparative color determinables, e.g. RED and YELLOW.

A diachronic phenomenal sorites series is one in which adjacent and nonadjacent color pairs are experienced at different times. If there were such a thing as a synchronic, phenomenal sorites series it would be one in which all of the colors are experienced simultaneously and would also be simultaneously experienced as bearing their various adjacency, nonadjacency, similarity, and nonsimilarity relations to each other. For ease of exposition, I shall often refer to these two kinds simply as synchronic series and diachronic series.

Diachronic series may come in two varieties. The first, where beginning and end elements are both of the same noncomparative determinable, like light blue, I shall call diachronic series with noncomparatively similar ends. The second variety, where begining and end elements are of different noncomparative determinables, like red and yellow, I shall call diachronic series with noncomparatively distinct ends.

Though I’ll raise doubts a little bit later, I’ll leave open for now whether synchronic series come in both varieties concerning the similarity or distinctness of the end members. I’m especially doubtful that there are synchronic series with noncomparatively distinct ends.

8.2. What the alleged problems are and how to solve them.
The four series kinds that I’ll be examining are:
(1) diachronic series with noncomparatively similar ends
(2) synchronic series with noncomparatively similar ends
(3) synchronic series with noncomparatively distinct ends
(4) diachronic series with noncomparatively distinct ends

8.2.1. Diachronic series with noncomparatively similar ends (diachronic blue/blue series)
Diachronic series with noncomparatively similar ends may be very small series. They may have as few as three elements. It’s highly unlikely that series with ends that differ in that one is red and the other is orange can be so small. Small series lend themselves to a certain ease of exposition, so they are nice to start with in explicating some of the main features relevant to discussing conceptualism and the intransitivity of indiscriminability.

It may seem clear, at least initially, that a diachronic phenomenal sorites series presents no real problem to conceptualism. In the example of the 3-item series, A and B look the same to conscious experience by my applying the same color concept to both. At some different time, B and C look the same by my applying a different concept than before to both. There’s no obvious problem that arises in hypothesizing B being conceptualized one way at one time and a different way at another time.

However, there are certain versions of conceptualism for which this sort of phenomenal sorites series does pose a problem. One way of interpreting the First Approximation as discussed in previous sections in connection with DIA is that the First Approximation embraces the following thesis concerning diachronic indiscriminability (at least for diachronic presentations of very short delay):

(DIASAMECON) If two colors are diachronically indistinguishable then the same concept is applied to each.

Now, in considering phenomenal sorites series, even of the sort that I am calling diachronic series, adjacent elements are indiscriminable not just diachronically: they are synchronically indiscriminable as well. One might naturally suppose that the kind of conceptualist attracted to DIASAMECON would also be attracted to the following thesis regarding synchronic indistinguishability.

(SYNSAMECON) If two colors are synchronically indistinguishable then, the same concept is applied to each.

But now we can work our way toward raising some serious problems for the conceptualist. Consider diachronic series with elements A, B, and C such that A and B are experienced at time t1, B and C and time t2, and A and C at t3. A and B are synchronically indiscriminable. And it is reasonable that any colors so similar as to be synchronically indiscriminable will also be diachronically indiscriminable. Suppose that A is experienced at t1 as blue. This will, according to the conceptualist, involve the application to A of the concept BLUE. In keeping with SYNSAMECON, at time t1, both A and B will be conceived of, color-wise, simply with the concept BLUE. In keeping with DIASAMECON, BLUE will also be applied to B at time t2. Similar appeals to SYNSAMECON and DIASAMECON will lead to the supposition that BLUE will be applied to C at both t2 and t3. But this looks to be a serious problem: at each time none of the colors is conceptualized with any color concept other than BLUE. On what conceptual basis can A and C seem different at t3?

Since, by hypothesis, A and C are discriminable, and A and C are experienced together at t3 and C is conceptualized as BLUE, then some concept other than BLUE will need to be applied to A at time t3. So A will be conceived of simply as BLUE at t1, and under some other concept or conceptualization at t3.

Now, it’s open to the conceptualist at this point to hypothesize that at t3, the other concept that is applied to A at t3 is an additional concept. That is, at t3, A is conceptualized under BLUE as well as some other concept, perhaps one comparing A to C so that the conceptual content at t3 involves something like A is a darker shade of blue than C. Given the initial supposition that, at t1, A was conceptualized simply as blue, we have it that A is conceptualized in two different ways at two different times. Now, the opponent of conceptualism may take it that there’s a slight air of implausibility in supposing that A is conceptualized in two different ways at two different times. But this is a minor problem. It’s not like the problem is an outright incoherence in the theory. To motivate that sort of accusation against conceptualism, it will help to turn to the next sort of series.

8.2.2. synchronic series with noncomparatively similar ends (synchronic blue/blue series)
Keeping our focus on a version of conceptualism like the First Approximation, we can see the problem that a synchronic series with noncomparatively similar ends poses. Sticking with the example of the three-element series ABC, we can see that the concept applied to A would have to be the same as the concept applied to B, and the concept applied to B would have to be the same as that applied to C. But this seems to lead directly to a contradiction in the theory, since, presumably it will want to account for the discriminablity of A and C in terms of a different concept being applied to each. To be clear, the point of this criticism is not to say that contradictory contents are being attributed to the perceiving subject. That is not so large a problem, for it is plausible that perceptual contents can represent things in a way that is necessarily false (as in certain illusions).[12] The problem here is that a contradiction is arising at the level of theory: it’s a contradictory theory of how perceptual consciousness works.
I think that we can motivate some serious questions about whether there can be synchronic phenomenal sorites series.

Consider, first, the question of whether there could be a series with very many elements, say 34 elements. Serious questions may be raised about whether foveal resolution and the capacity of attention genuinely allow for all 34 elements and their various relevant relations to enter into conscious experience all at once. It is one thing to stick all 34 colors up in front of someone’s face synchronically, but the limitations imposed by overt and covert attention may force the colors and their relations to be taken in diachronically after all. The subject may be restricted to moving a limited window of attention across the spatial array and taking in various color pairs diachronically. There may thus be no color that simultaneously looks just like two manifestly distinct non-adjacent colors.

The natural suggestion, of course, is for the nonconceptualist to suggest the existence of a small series. With only three elements, it is much more plausible that all three colors may be taken in all at once. This would make it more plausible that the relevant similarities and differences are taken in at the same time. Note, however, that for a very small series, the nonadjacent colors won’t be very different. They will be nowhere near as different as unique red and unique yellow, or even as different as red and orange. It would be puzzling to say of a color that it simultaneously looked just like red and just like yellow. It’s puzzling because of how different red and yellow look. But if A and C look very similar to start with, it’s not obvious that it’s so problematic for B to be conceived of as simultaneously looking like A and like C.

Note that in the previous paragraph I said that the 3-item series is “more plausible” to regard as synchronic. But this is not to concede that it actually is plausible. With very similar color pairs, such as the ones in the figures 1 and 2, it takes some non-negligible amount of time and attention to see the difference between the two. Such considerations may be recruited to help raise doubts about whether even the smallest phenomenal sorites series is small enough to be synchronic.

Another move available to the conceptualist is to exploit the sort of indeterminacy invoked earlier in discussion of DIA. Thus, the conceptualization of A and the conceptualization of B will each be noncommittal as to which maximally determinate shade of, say, light blue, A is and B is. Such indeterminate contents will be consistent with A and B being the same determinate shade and also be consistent with A and B being distinct determinate shades of the same determinable.

8.2.3. synchronic series with noncomparatively distinct ends (synchronic red/yellow series)
Such a series would have to be larger than a 3 element series. It is quite implausible that there could be a phenomenal sorites series with endpoints differing as much as a red-yellow difference or even a red-orange difference that had as few as only 3 elements. And the larger the series, the less plausible it is that it could be a synchronic series.

8.2.4. diachronic series with noncomparatively distinct ends (diachronic red/yellow series)
Let such a series have a beginning element that is unique red and an ending element that is unique yellow. It is highly plausible that the concept applied in experience of the first element will be RED and not YELLOW, and for the last element, YELLOW and not RED.
Now consider what we can call “a forced march” through a diachronic phenomenal sorites series wherein colors are presented one at a time. If the delays between color presentations are shorter than the term of the memory buffer, then it seems tempting, at least to the adherent of the First Approximation, to say that diachronic indistinguishabilty is going to need to be explicated by sameness of representation. This is what adherence to principles like DIASAMECON require. However, here’s where a problem arises: each member of a pair of adjacents, for all adjacents in the series, is diachronically indistinguishable from its neighbor, and thus what’s conceptualized as red at the start of the march is going to lead to a RED conceptualization of unique yellow at the end of the march. But this contradicts the previous hypothesis that the end element would be conceptualized with YELLOW and not RED.
So, there’s going to be some non-end element that is conceptualized in different ways at different times. At this point, the conceptualist can argue that this can be made plausible as a context effect where what counts as context may include what Raffman (1996) calls internal context: differences in what concepts are applied to a presented color are due not just to what else is currently presented, but also to different internal states that reflect the recent history of having been “marched” through the series in one direction rather than another.

This general strategy, which countenances changes of what concept is being applied to a given color in the series, is especially problematic for the First Approximation. Central to the First Approximation was the thought that indiscriminable shades would be conceptualized in the same way. I’m not going to dwell here on problems for the First Approximation, for we’ve seen other reasons to abandon it.

The sort of phenomenal sorites series currently contemplated may be seen as raising certain problems for the Second Approximation. Since there will need to be a change in the concepts deployed at some point in the series, one might wonder whether such a mid-march concept change count as a kind of forgetting and, if so, count as a violation of thesis (M) relating concepts to memory.
It seems that the defender of the Second Approximation has some promising responses at this point. One is to consider this forgetting as tolerable and no threat to the present form of conceptualism. The forgetting may be regarded as due to a kind of interference. Further, such interference effect can be regarded as consistent with (M) since (M) is an empirical generalization, not an analytic constraint on the concept of a concept.

Of course, one can easily remember that one said that the last chip was red and not orange, but the forgetting that is relevant involves not what concept was deployed in speech, but what concept was deployed in perception. When one arrives at the first chip in a march to be conceptualized as orange instead of red, one’s confidence falters regarding what the previous chip looked like.

8.2.5. Summary of remarks about phenomenal sorites
The main points of the preceding discussion of phenomenal sorites are the following. First, there are four kinds of phenomenal sorites series which differ in part with respect to how serious of a problem they seem to pose to conceptualism. Second, for all four kinds, the conceptualist has responses at hand for dealing with the alleged problems. For the large synchronic series, the conceptualist can plausibly deny the existence of such series. For the large synchronic series, the conceptualist can make a plausible case that the concepts applied shift during a “forced march” in such a way as to count as a kind of memory failure. Given the rejection of the Re-identification constraint as an a priori constraint on concept possession, such memory failure need not pose a threat to the conceptualism on offer. For small phenomenal series, the conceptualism on offer can accommodate such series by appeal to the indeterminacy of the relevant concepts.

9. Conclusion
I’ve argued for the viability, in the face of worries about fineness of grain, for a conceptualism about consciousness of colors that does not lean on demonstrative concepts. Central to the treatment that I favor—what I’ve called the Second Approximation—is to emphasize the indeterminate content of many of our color concepts. Also key is regarding the relation between memory and concepts as an empirical generalization, not as an analytic component of the very idea of a concept.

A much shorter early version of this paper was presented at the Second Consciousness Online conference in February of 2010. Jake Berger, Philippe Chuard, Charlie Pelling, and David Pereplyotchik presented highly detailed an useful commentaries there for which I am enormously grateful. I am grateful too for helpful and interesting comments from Richard Brown, James Dow, Aspasia Kanellou, Michal Klincewicz, David Rosenthal, and Josh Weisberg.


[12] I don’t mind supposing that reality has no room for contradictions. Something cannot at one at the same time be just like A and not just like A. But it’s much less problematic allowing that there are contradictory representations. There is, for example, the following sentence: “B is a color that is simultaneously just like A and not just like A.” That sentence gets on just fine being contradictory. Perhaps analogous mental representations exist while being analogously contradictory. Of course, when the representations in question are beliefs, and the believers are rational, and the contradictions are very simple and obvious, many philosophers will want to say that there’s some sort of problem here. But the conceptualism on offer is not committed to conscious experiences being beliefs. Conscious experiences need only be similar to beliefs in the following manner: they are attitudes toward contents exhausted by deployed concepts.