I have to say, I think Luke made a rather cogent argument. I'm glad you posted that article by Sam. It coincides perfectly with my discussion on functionality toward the beginning of my response. Nevertheless, Harris' response blunders in trying to say that because we share certain perspectives on something we can have a science of it. He uses medicine, as I did on my blog (along with things like, say, engineering). Medicine certainly is an ends-oriented enterprise, but science plays a role. However, Harris' central thesis is that science can determine those ends. That is not at all how medicine operates. Science informs us about those functions on which we've fixed our perspective. We can say what is good for the human body and analyze the human body scientifically to best determine how to achieve those ends that we've determined to be important. That is the important distinction: the science behind medicine did not determine those outcomes; though, certainly, scientific descriptions and clarity help us refine and perceive more clearly the ends we that want to achieve.

Returning to Luke's argument, though, it is not entirely unlike my central confusion. Harris is defining morality in one way (brain states) while defending it from another (optimizing well-being). Luke's argument adequately demonstrates how Harris claims many things that are either vacuous or tautological/circular. As he says, "A great many moral philosophers would agree that if we define morality in terms of the “well-being of conscious creatures,” then obviously most of moral theory becomes a science." The position is entirely contentious for the fact Harris leaves "well-being" on a sliding scale. For instance, he effectively equates it to the improvement of certain brain states. Harris claims that "morality ... really relates to the intentions and behaviors that affect the well-being of conscious creatures." HOW does this relation operate? If well-being is defined by the advancement of certain values that are in turn characterized by certain brain states, how does someone dying to promote an end count as well-being? Certainly the person dying has lost well-being. But this is an odd slight of hand to relate well-being to brain states and then say Jon's action improved Jan's brain state, therefore well-being has been improved. Well, not quite. This is a blunder of classical utility theory. Harris' conception is nothing more than a vague notion of redefining pleasure/utility in terms of neurobiology. It does not eliminate the problems of utility qua pleasure fulfillment. This is where Luke's initial statements are most apt. This is also why my response relying on Amartya Sen's work challenging the utility framework that has been systemic on social theories plays a role.

Since I wrote my blog last night, I decided to re-read the end of Sen's The Standard of Living on my way to work today (and consequently have the book with me). An example from his closing summary has import here. One of the other writers (Ravi Kunbar) in this lecture series talks about the living standard (a part of well-being) by contrasting how we divide "the cake" up between two hypothetical people, Sen and Williams. Kunbar argues that "taking some cake away from Sen and giving it to Williams ... would not be much persuaded by arguments that such an arrangement would in fact lead to catastrophic reduction in the standard living of both." To which Sen replies, "That is certainly right, but there is no question that taking the cake away from 'Sen' would reduce his standard of living," (106) The context of this hypothetical situation is certainly different, but it deals with a fundamentally similar problem: the concept of the living standard cannot be argued for in one way by its aggregate component when what is at stake is also the impacts on the constituents of that aggregation. It is a composition fallacy to think there is a direct (one-to-one) correspondence between the two that we can simply ignore, in this case, the welfare decrease to 'Sen' when there is no such decrease to the joint 'Sen & Williams.' Unacquainted with this topic you might think, "of course the aggregate can be just as the term implies: a summation of the ups and downs of the people in the population." That is only one way to characterize our aggregate. That is to say, Welfare[population] = Person1 + Person2 + ... + PersonN, for some N people in the population. But recognize the implicit coefficients on each 'Person'. They each have an unstated coefficient equal to unity. In reality, as long as we assume Welfare is aggregated linearly, the form is really Welfare[population] = A*Person1 + B*Person2 + ... + Z*PersonN, where A = B = ... = Z = 1 in this case. Why should it be this way? Why should it not be some other way, such that some of the sub-population ought to be weighted differently in aggregation. There are very good reasons to alter this, just like we use a progressive tax system to accommodate for the fact some people are made comparatively worse off by taxation than more affluent parts of the population. In terms of welfare, some people are physically unable to benefit from the same commodity baskets other people enjoy, whether it be due to a nutritional necessity or maybe a physical handicap, and this may obviously change over the course of each person's life. It may seem odd to throw nutritional needs into a welfare or utility assessment, but this idea dates well back to centuries ago (e.g., Lagrange, I believe Sen mentions in Ethics and Economics).

How does this apply to Harris' broad concept of well-being? I think it is very analogous to what I just explained. Consider these two cases.

(A) Jon sacrifices his health (maybe his life) to achieve his dream. In doing so, many people in the future will be made better off.

(B) Jon dies so that Jan can live out the rest of her life.

Barring issues about uncertainty of outcomes, there is something significantly important in how we handle these two situations. (A) is mentioned in my response. Notice that Jon's welfare is decreased just as hypothetical 'Sen' had a welfare decrease. The same applies to (B). In this instance, we're concerned with the brain states on Harris' account. In (B) we can most certainly say Jon has lost everything. How do we aggregate Welfare[Jon & Jan]? Certainly Welfare[Jon] = 0 now, and Welfare[Jan] at this time is greater than it otherwise would have been, but how do we compare Welfare[Jon & Jan] under the conditions of (B) to some other condition? This goes right into the issues in dimensionality I discuss toward the end of my blog. This is like trying to compare the parameter space values (5, 7) with (7, 5). I used the example from your list where we're looking at (self-respect, personal liberty). In this case, we want to know the parameters of welfare by people (Jon, Jan). How do we weight (0, 1) with (1, 0)? The same issue is implicit in the living standard example and in the two cases I've presented.

The fact is that Harris wants us to believe we can draw the kind of moral landscape he shows on his TedTalk. I find that entirely begging the question. I had thought I read it in the book this morning, so that I could use the quote; it turns out that I must have read it in chapter two of Ethics and Economics before bed last night. The point was iterated in my response already. I'll reiterate it here. I will also point out two other issues: one formal and the other practical. Going back to the "landscape" picture Harris presents we can think of it as our regular 3-dimensional coordinate system. The ground-level is the xy-plane and anything in the +z direction (toward the peaks) is a moral improvement. How do we characterize this system (x, y, z)? That is fundamentally the issue. Now the (x, y) space is unimportant. Maybe it's location and person. What is of consequence is that z is a composite indicator that unifies all moral value into one metric. One might say that we can really break z apart and talk about higher dimensional spaces, but this still begs the question (1) how do we weight the values in the z-space given the discussion above? (2) How do we know there is a "peak"? The idea of a peak is an optimization, but in the case presented there is only one direction we can go to reach it. Imagine for the moment that one could go in either the +z or +y direction for moral advancement instead. Now where do we identify peaks? One might be moving more z-ward and less y-ward. Is this an improvement? If we cannot distinguish those sort of changes precisely, then the moral metric Harris wants us to believe in "must" be well-being cannot be substantiated.

Another thing I did not discuss on my blog steps away from the availability of the optimization framework. A lot of Sen's earlier work focused on utility preference theory and its possible alternative formulations (see, for instance, his Choice, Welfare, and Measurement). Utility theory works in modern economics by making many gross assumptions about preferences that allow us to handle some rather complex utility or production functions in terms of optimization (differential) calculus. These assumptions characterize the axioms of consumer choice theory. They include things like transitivity: A is preferred to B is preferred to C implies that A is preferred to B. It also includes things like if A is preferred to B in one choice set (say, a set consisting of just A and B) that is is preferred in all choice sets (say, a set consisting of A, B, and C). Assumptions like these are false in reality. Behavioral economics has demonstrated that the choice bundle does impact consumer preference. Dan Ariely has a couple TedTalks that demonstrate this. For instance, choosing between a magazine subscription options (electronic, print) produces difference choices compared to them made under (electronic, print, both). The reason we need assumptions like this is so that we can order things monotonically (in a straight-line, like the z-axis). If we cannot do this, we cannot talk about optimization. We can have kinks and bends and branches in our ordering that never meet up again. In order theory, these are called partial orders. If Harris wants to talk seriously about morality in the terms he is using and the picture he wants to paint, it is more realistic that morality is dominated by partial orders not amenable to the type of optimization theory he would have us believe exists. Thus, not only does he face issues under an optimization calculus, he may not even get that much.

A practical example associated with my first criticism above is expounded in Mismeasuring Our Lives for which Sen is a co-author of the report that comprises the book. In it the authors (probably Sen, among others) argue how an indicator of welfare (or environmental fitness, or anything for that matter) cannot be summarized into one measure. For a long time organizations like the World Bank or the IMF characterized any measure of success--whether it be economic, social, whatever--entirely in terms of GDP. This has posed a lot of problems for a long time. Instead, the authors insist that we need multiple measures that accommodate the various factors involved. The use an "instrument panel" analogy. You would not be served well by having one gauge that tells you about both your fuel and speed. This indicator, while meaningful and useful in some capacity, would serve entirely problematic. Instead, we have separate indicators that inform us about our fuel level and a separate one for our speed. By looking at the two we can better manage our driving behavior, by getting gas when we need to or slowing down when it is appropriate. The unified indicator may still get us to the right conclusion, but probably not, certainly not effectively or consistently. This is an extension of research into environmental sustainability indicators and welfare indicators as the work on the living standard presents.

I think this practical example demonstrates the situation we're in quite nicely, since I think these sort of social scientific situations are the best manifestation of moral theory we can ever face. Harris wants us to believe we can have a science of morality. I do not disagree that science can play a role, but it cannot determine the ends of our moral inquiry. It certainly cannot do it in the unified capacity for which Harris argues as the basis of moral values and, consequently, of well-being. He is ultimately trying to advance a shoddy utility theory that doesn't overcome any of the problems that utility theory faces. Instead, science can help us determine what indicators we can have available, how to obtain them, and to assemble them. The "instrument panel" view asserted above is exactly how science plays its role in public policy, which is exactly one of those situations where science does aid our moral quests in life. But science will never determine how we read that instrument panel or ultimately decide which instruments we place on it. No amount of understanding the brain will change that. It will only improve those indicators by improving our information base, and consequently, our ability to choose among possible indicators (e.g., we used GDP for so long because it was particularly easy to obtain, but with advancement in our methodology and available information there is no respectable reason not to expand our use of indicators). I have focused in this comment on elaborating that positive side of Harris' argument while also demonstrating how his conception of morality does not apply. That is why I say he got morality wrong, and why I think Luke is right in his assessment on whether Harris has a positive argument for this conception. There are far more ethical arguments and meta-ethical arguments I could make, but they are not as important. They show up in my response, but I try not to emphasize them too much. The point is that there are even graver problems lurking behind the scene.

In closing, I just want to say that I think Luke is wrong. The cover of Harris' book is nice. It has a flashy minimalist theme, using two colors and a gradient background. I like minimalism! Furthermore, I think his idea of desirism fails, which extends directly from one simple fact (among others): desire-independent reasons for action exist. For more details on that I suggest reading John Searle's work in Intentionality or his latest book, Making the Social World. I make use of his decades of research in my latest paper.