Discussion:
  1. Arno G. Wouters (2007). Design Explanation: Determining the Constraints on What Can Be Alive. [REVIEW] Erkenntnis 67 (1):65-80.
    This paper is concerned with reasonings that purport to explain why certain organisms have certain traits by showing that their actual design is better than contrasting designs. Biologists call such reasonings ‘functional explanations’. To avoid confusion with other uses of that phrase, I call them ‘design explanations’. This paper discusses the structure of design explanations and how they contribute to scientific understanding. Design explanations are contrastive and often compare real organisms to hypothetical organisms that cannot possibly exist. They are not (...)
    Direct download (6 more)  
     
    My bibliography  
     
    Export citation  
Back    All discussions

2009-12-24
Can we make room for developmental constraints?
Design explanations are explanations, or maybe just arguments, addressing questions about why certain organisms have some traits instead of others. For example, since tetrapods have lungs but don't have gills, it seems reasonable to ask why. Design explanations attempt to answer such questions by looking at functional dependencies and integration between different traits in the same organism. For example, we might start by looking at the functional requirements for respiration in a large organism living on land, invoke the relevant laws from physics or chemistry or biology, and show that having gills would make the organism less viable. 

Wouters proposes a schema for design explanations. In my words:

1) Specify the organism's properties and conditions of existence.
2) Assert that trait T possessed by the organism is more useful than alternative trait T'.
3) Provide an explanation of 2).

I see 2) as an undue limitation. Contrasting alternative traits is a very important strategy but we could make it more flexible. An additional strategy for design explanations could be:

A) Answer a question of the form: "why no members of class X possess trait T'?"
B) Replace trait T' with T and show that the answer is not valid anymore.

Following this strategy we could keep Wouters's schema by replacing 2) with:
2') Assert that trait T' is either "impossible" or would make the organism less viable.
Here, "less viable" could also be construed as a disruptive effect on the organism's integration (see below for more), and does not necessarily involve comparison with another trait. 

Why should we want to make this change? Let me remind that we are not trying to devise a new kind of explanation, but to capture what biologists do when they give explanations. I'm sympathetic to Wouters's claim that design explanation "can be understood from a mechanistic point of view as revealing the constraints on what mechanisms can be alive". So I wonder why we shouldn't try to make design explanations more relevant to the discussion on biological constraints.

Constraints operate at the level of developmental mechanisms and prevent the occurrence of certain types of variations. For example, while there is variation in the number of segments of centipedes, no centipede has an even number of segments. This kind of constraint does not arise from the fact that organisms possessing a certain trait are less viable, but presumably from the "logic" imposed by some mechanism over the process of segment formation during development. For example (just a speculation), one could imagine a first "unconstrained" process of segment formation along the main axis, then a single division of each existing segment into two, except for the "head" segment.

I said "biological constraints", but a better way to frame the discussion would be in terms of a taxonomy of processes operating at the level of phenotypic evolution. To keep things short I will refer to a paper by Giuseppe Fusco, where in addition to developmental constraints the notion of "internal selection" is defined and defended. Briefly, internal selection affects the viability of an organism at the level of integration. The main difference with natural selection is that the effects of the former "remain approximately constant across a wide range of environments" (Fusco 2001).

The change I am advocating would make it possible to include into a design explanation those specific (i.e., less generic) constraints that may prevent the production of alternative forms. My point is that these alternative "impossible" forms are already part of the explananda of evo-devo research and can be assumed to have a mechanistic explanation, so they would fit in well with an enlarged schema for design explanation. 



2010-01-31
Can we make room for developmental constraints?
On p. 78 of my paper I mention three kinds of limitations on living beings' design: 
  1. it must be possible to generate individuals with that design
  2. once generated, a design must be viable
  3. there must be an evolutionary path to that design
Let's call them: (1) constraints on what mechanism can be generated (generative constraints), (2) constraints on what mechanism can be alive (functional constraints), (3) constraints on what mechanisms can evolve (historical constraints). I think it is very important not to confuse these three different type of constraints. Design explanations, as I use that term, are concerned with functional constraints. I am concerned that broadening the notion of design explanation such that it includes appeals to other types of constraints would easily lead to confusion of generative constraints and functional constraints.

Concerning developmental constraints I would think that they come in two kinds: functional developmental constraints and generative developmental constraints. If a certain trait does not develop because it has a disruptive effect on the integration,  the lack of that trait is the result of a functional developmental constraint (such a trait is possible (it can be generated), but organisms with that trait would not be viable). If a certain trait does not develop because it does not fit into the logic of a developmental mechanism it is a generative development constraint, such a trait is impossible (but it could be that if it were possible the resulting organism would be viable).






2010-05-25
Can we make room for developmental constraints?
Long after the original post I find myself moved to raise two questions. The reason is that my own field is not biological evolution, and there were two concepts here introduced that I am trying to reconcile with their use in other fields.

The first of these is the notion of "constraint". In the position outlined by Arno Wouters, it seems apparent that he employs the term somewhat in the sense that a constraint blocks degrees of freedom. However, as an outsider I would have assumed that when it comes to biological evolution we speak instead of probabilistic processes. That is, instead of the probability of a morphological change having a probability of 0 or 1, it would be a probability between 0 and 1, where that of 0 and 1 are only hypothetical limiting cases. The former is an unequivocal determinism; the latter, a probabilistic determination. So I have a question: do biological evolutionists recognize this difference, and do some employ the latter? What lies behind this question is the challenge of offering an explanation of emergent phenomena rather than mere description (supervenience).

The other concept is "function". This also I find difficult to reconcile with the extensive discussion of functional properties or of structural functionalism in other fields. I don't think anyone would have a problem with a statement such as this: the structure of a non-equilibrium system (such as a biological organism) requires that certain conditions be met, primarily a metabolism, but also the presence of certain subsystems the behavior of which is a necessary condition for the existence of the structure of the whole.

Now, as description, this seems a natural way to characterize a state of affairs and does not seem problematic, but as explanation it has for decades been subject to such criticism that people now hesitate to employ it that way. Let me see if off the top of my head I can briefly say why.

a) A "functional" relation of entities implies that the relation of their observable properties is probable and for that reason their relation is stable.  This may suit subsystems, but when it comes to the organism as a whole, it is widely assumed that it is an emergent process, the properties of which by definition cannot be accounted for in terms of a Laplacean knowledge of its parts. That is, if we assume that organisms have unpredictable unique properties that are not unequivocally determined, how then can the relation to its constituent subsystems, its "base level", be characterized as probable or stable? A relation of dependence is not necessarily reducible to a probable relation of observbables.

b) A functional property of an organ is an unobservable property that is inferred from its satisfying a need of the whole. Again, this seems more descriptive than explanatory, for if the behavior or properties of the whole emerges as a result of its constituents, how can this outcome also be a starting condition? That is, functionalism, as explanation, is accused of being circular and teleological.

c) Another criticism of functionalism is that if a system consists of subsystems that have a functional relation with the whole, then the resulting system becomes necessarily static, incapable of change except in response to outside contingencies and so is passive rather than active. I assume that activity or agency is fundamental to what we mean by life, such as modifying behavior to ensure continued metabolism. To put it rather strongly, adaptation seems a kind of death where your relation with your environment settles to what is most probable, and this is dust to dust.

Now, these are criticisms of functionalism raised by people who experience frustration in their search of explanation and are not satisfied by mere description. And so my question, if you will allow, is whether biologists generally seek explanation or does description serve as a kind of  explanation? 



2010-05-25
Can we make room for developmental constraints?
Reply to Haines Brown
Haines, I find it difficult to relate to your question, because I have no clue about what you are talking about and what you are asking. This is no doubt because of differences in our background. I will try to explain some of the things I do not understand.
(1) You write:
In the position outlined by Arno Wouters, it seems apparent that he employs the term somewhat in the sense that a constraint blocks degrees of freedom. However, as an outsider I would have assumed that when it comes to biological evolution we speak instead of probabilistic processes. That is, instead of the probability of a morphological change having a probability of 0 or 1, it would be a probability between 0 and 1, where that of 0 and 1 are only hypothetical limiting cases. The former is an unequivocal determinism; the latter, a probabilistic determination.
I am not clear about your notion of degrees of freedom and I am not sure that my notion of constraints is one of blocking degrees of freedom. However I am very surprised that you think that my talk of constraints is at odds with a view of evolution as a probabilistic process. In my view, the constraints on being alive are independent of the kind of process by which organisms are generated, deterministic, probabilistic, emergent evolution or whatever, and I have no opinion about the question which of these processes are able to generate living systems and which not.

(2) You write "A "functional" relation of entities implies that the relation of their observable properties is probable and for that reason their relation is stable."  and "A relation of dependence is not necessarily reducible to a probable relation of observables."

I have no idea what you are talking about. Can you clarify your point by means of an example of a functional dependency from my paper, such as  'living an active life on land functionally depends on having lungs rather than gills'?


 (3) I don't understand your talk of functional properties as unobservables. Actually, I almost never understand philosophical talk of functional properties, but talking of functional properties as unobservable makes it especially difficult. 

I understand biologists when they talk of functional properties as opposed to morphological properties: morphological properties concern the shape and structure of parts or wholes, functional properties characterize activities and processes. Both kind of properties are observable and often measurable. 

I also understand biologists when they talk of the function of hearts and lungs, etc.. In my view such talk refers to roles in the generation of an activity of the whole. In this context the term 'functional property' means something like 'a property that enables an organ to perform its role in the whole". In this sense the property of the heart to have a four chambered structure is a functional property and its property to produce heart-sound not (except for some mammals living in modern western countries). However, both the four-chambered structure and the production of heart-sounds are IMHO observable.

Please, see my "Four Notions of Function" (Studies in History and Philosophy of Biology and Biomedical Science 34(4): 633-668) for my views on function talk in biology.


(4) You write:
 if a system consists of subsystems that have a functional relation with the whole, then the resulting system becomes necessarily static, incapable of change except in response to outside contingencies and so is passive rather than active.
and
adaptation seems a kind of death where your relation with your environment settles to what is most probable, and this is dust to dust.
I have no idea about what you may mean. Why would a system whose parts have functions in bringing about the activity of the whole be static and incapable of change? What do you mean by 'outside contingencies'? What by' passive' and 'active'? What by 'adaptation'? 



(6) You say: "Now, these are criticisms of functionalism raised by people who experience frustration in their search of explanation and are not satisfied by mere description"

It will be clear from what I said above that I don't understand these criticisms, but this remark makes me very curious to learn more. Can you refer to a couple of papers? Or people? 

Can you explain in somewhat more detail how those criticism apply to my view that design explanations are non-causal explanations?


(5) You ask:
And so my question, if you will allow, is whether biologists generally seek explanation or does description serve as a kind of  explanation? 
I think that biologists seek explanations, as do all scientists. I doubt that there ever was a time when biologists merely looked for descriptions, but if there was such a time it was long ago.

2010-05-26
Can we make room for developmental constraints?
Reply to Arno Wouters
1. I borrowed, perhaps a bit adventurously, the notion of degrees of freedom from the definition of state space in physics: the number of variables needed to define the state of a system. This seems an epistemological definition because it reduces  the world to one's observations and inferences drawn from them (and one remains agnostic about anything else). I believe the other notion of constraint is ontological and constraint is a structure that determines the relative probability of possible states. They seem to differ profoundly in their implications, and hence my concern for which definition you had in mind. The probabilistic constraint does not presuppose what the possibilities are, which are a priori and are basically not entirely known, but the constraint increases the probabily of some possibilities over others.. Of course, one can infer from outcomes what a subset of the original possibilities were, but this is not the aim. The aim is not to predict, but to assess the effort (work) needed to actualize a possibiity defined by a probability distribution arising from actual structural constraints on possibilities  (this gets complicated, but perhaps this is a hint). It seems this makes a lot more sense in evolution, where emergent properties appear that cannot be reduced to the observables of a base level, but instead actualize possibilities.

2. Sorry to distress you with my verbal swamp. Functional properties, functionalism, and structural functionalism are much discussed and criticized. This criticism (taking place largely in the social sciences), amounts, in my view, to three points: a) Functionalism presumes there is a probable relation of entities in that the functional role of one depends on specific properties or behavior that satisfies a need of the other. This tends to lock in those properties and so reduces a system to a functional coherence (maximum entropy) that precludes change, when in fact all systems do change. b) Functionalism is teleological or circular in that an emergent whole arises from the functional properties of its constituents and yet this functionality is their result. c)  Functionalism describes a relation of properties, such as proximity or constant conjunction, but is not an explanation. There's no agency that somehow works contrary to entopy increase. 

To demonstrate the distress caused by problems of functionality, I happened to be reading this morning, John Searle, The Construction of Social Reality (I'm not recommending the book). He says that function depends on subjective values, and so one cannot say the heart functions to pump blood without adding to a causal relation a subjective value judgement that it's good to be alive. Most people, including you and I, have no problem saying that  lungs perform a functional for land animals, but I believe such a statement is incompatible with closed systems. Basically, living organisms are far-from-equilibrium systems that depend on certain conditions to maintain their emergent structure. For example, without metabolism (other than viruses, I suppose) the structure necessarily deconstructs (death) without it. I therefore have no problem speaking of animals or humans as having "needs", the satisfaction of which is necessary for life. Logically, A functional organ,  property or behavior satisfies needs so that life to continues, which is a Telos which I assume is hard-wired in most living organisms because a property giving rise to persistence seems more likely to prevail over other combinations of organic molecules that lack it.   

 3. Your "Four notions" article is not accessible in the journals database I use, and so I'm put into the unfortunate position of commenting on your remark without having read it yet. However, I don't believe the problems here are particuarly difficult. True, the term "unobservable" lacks a consensus definition. My approach to this is to define it by negation. An observable is a property of a system state that can be determined by a physical operation. So I like to define "unobservable" as that which which there is no such operation and thus no frame, is not localized by the relation of observer and observed. Now, a functional property is extrinsic, a relation of a part to the whole, and is therefore not framed, but inferred from an observed relationship.  A morphological property is a property within a frame, what you call a whole. However, I don't know why an activity  becomes unobservable. The heart beats, and this activity is observable; an organism is alive, and we know this from observing its behavior. Now, process seems a bit more different than activity, for like St. Augustine's remark about knowing what time is but being unable to describe it to others, process is intuitive and so far there is little progress analysing it terms that make the concept operational (something I've been working on). Although I might be pillored for saying so, the person who comes closest to it seems to be Heidegger, in Sein und Zeit. Nevertheless, I believe we basically agree here.

4. I addressed before the question you here raise. In terms of thermodynamics, all "closed" systems move toward a more probable state. For example, when a crystal forms in a supersaturated solution, the molecular bonds represent a more probable state for the molecules than remaining unbound in solution, and the bond is the probable relation of molecular properties. The result is novel (a crystal, because of the bonds, holds more information than the sum of individual molecules), but it is predictable from a knowledge of molecular properties. The word emergence originally (basically defined in a 1926 article by Pepper) meant both novelty and unpredictability, but in recent years it has come to mean just unpredictability. A closed system is predictable because its behavior is determined by local properties. My intent was not to suggest that systems having functional parts are static, but that speaking of functionality only makes sense for emergent and therefore open systems, such as living organisms, and if the system is hard-wired with a Telos to perpetuate itself (a tropical cyclone is an emergent system that is open and has a "metabolism" (exothermic evaporation of rising surface water), but is not hard-wired with a Telos to perpetuate itself. It collapses when the conditions that supported its structure change. Specifically, the Coriolis Effect brings it over cooler waters. This is what I meant by a "contingency": a factor that influences the system but is not defined so as to fall within its frame; it is an "outside" and thus "accidental" determination. If we were to enlarge the frame to bring in the Coriolis Effect, then the tropical cyclone acquires a probable life span rather than simply go on indefinately, but the Coriolis is no longer a contingency, but part of the system under study.

A closed system may have Gibbs free energy and so change, but only so long as that energy has not dissipated. Ultimately, the notion of such a causal power begs the question: how did the closed system acquire its potency, and we enter an almost infinite regress. Such a closed system is doomed to move toward its most probable and stable state of maximum entropy, although in some cases it might require cosmic time to get there. An open system is not so called because it is open to outside influences, which depends on our subjective frame, our definition of what is inside and outside. Rather, an open system seems to be one that is emergent in the sense that it moves toward a less probable and therefore a less predictable outcome; it is far non-equilibrium, and must sustain its improbable structure by a continual dissipation of energy, as in the cyclone example above or in living organisms or an eddy in a rushing stream. These systems have properties and behavior that do not reduce to the properties of its parts.  My point about passive/active is that in the former, change in the system reduces to local observable properties and the influence of contingencies; in the latter, the system escapes causal determination, which is why it is called emergent, and because emergent properties are not reducible, can alter its own state in a way that is not simply adaptive. A squirrel gathering nuts in preparation for the coming winter is not responding the pressure of an immediate need, but to what it knows will be a coming need. But I'm no biologist, and so can only speculate.

6. I'll comply with your request, but first I should say that "design" explanation leaves me mystified, in part because the begining of this thread does not seem to be archived and so I lack context. My dictionary says the verb "design"  is the product of human intentionality, and the noun design is the result of such intentionality. If design is a property that is independent of consciousness, then I'm mystified, but that may well not have been your intent. If "design" is a notion tied to human intentionality, then a simple view would be that the mental concept was the cause of resulting design, such as in architecture. However, I have a perverse view that bypasses causal explanation, so that I don't have to conclude that the mind has some kind of mysterious causal power that can produce enormous physical results depite the Second Law of Thermodynamics, getting something for nothing in terms of energy. Rather, I see the relation of the designer and the object designed as what I call a "superposition of processes" (no connection with QM intended). This notion is very complicated, and I'm sure you don't want to hear about it, but the bottom line is that the designer's thought shapes the design without having actually caused it.

a. Important critique, although rather old: Robert Cummins. "Functional Analysis." The Journal of Philosophy 2 (1972), 741-765 
b. A newer critique, although one I find worrisome: J. Holmwood. "Functionalism and its Critics." Modern Social Theory: an Introduction, ed. A. Harrington (Oxford: OUP, 2005), 87-109.
c. A point I use: causal mechanism can be used for functional explanation, but if included is more than just intrinsic properties: Daniel Steel. "Mechanisms and Functional Hypotheses in Social Science."
    Philosophy of Science 72 (2005), 941-952
d. Useful for broad overview of Durkheimian tradition and rise of neofunctionalism: Alexandra Maryanski and Jonathan H. Turner. "The Offspring of Functionalism: French and British Structuralism."
    Sociological Theory  9 (1991), 106-105.
e. Another old standard critique: K. Davis. "The Myth of Functional Analysis as a Special Method in Sociology and Anthropology." American Sociological Review 24 (1959), 757-772.
f. A useful analysis of functionalism's fall, which sheds light on my concern over closed systems: S. N. Eisenstadt. "Functional Analysis in Anthropology and Sociology: An Interpretative Essay."
  Annual Review of Anthropology 19 (1990), 243-260.
g. For my point that functional analysis is descriptive rather than explanatory: Harold Fallding. "Functional Analysis in Sociology." American Sociological Review 28 (1963), 5-13.
h. For functionalism and mind from philosphical perspective: Alan Weir. : "More Trouble for Functionalism." Proceedings of the Aristotelian Society N.S. 101 (2001), 267-293.
i.  A more sympathetic, although dated, philosophical approach:  Carl G. Hempel. "The Logic of Functional Analysis." Symposium on Sociological Theory, ed. Lleyellyn Gross (Harper&Row, 1959), 271-307.
j. Usefully raises issue of meaning of functionalism in different domans: Mahner and Mario Bunge. "Function and Functionalism: A Synthetic Perspective." Philosophy of Science 68 (2001), 75-94.

Well, this is a disordered list and done quickly, but it may have some use. There's many old titles because after the 1970s, functionalism was (arguably) a dead horse. I'm curious how biology relates to these kinds of issues.
 
5. Ah, a comforting thought. In historiography, the dominant paradigm is the view that any explanation ("grand narrative") that does not reduce to epistemology (such as language) is impossible. In sociology,
   although there are some old-time exceptions like Eisenstadt above, there is a similar move to marginalize explanation implicitly. I haven't seen that in the natural sciences, although I suspect antirealism
   in the philosophy of science is undercuts explanation. As for biological evolution, I've no doubt there is explanation, perhaps what is referred to loosely as Darwinism, that is arguably not unlike that of
   other historical science (people seem widely to assume that the Darwinist paradigm is widely embraced, but I'm not convinced of it. But it does seem to gives a different meaning to the word "explanation".
   Speaking loosely and in traditional terms, doesn't "explanation" mean that there's a "necessary" relation between one thing an another and, versus mere description of that relation, there's something
   compelling or determinant about it? I suppose this compulsion in positivist terms refers to inflexible universal laws, the compelling nature of deductive logic, perhaps Kant's rationality, and Newtonian
   causality. This seems ill-suited for explanation in the historical sciences (I go so far to argue that causal explanaction is a limiting or special case in all scientific domains), and yet explanation looks to
   a relation, to what makes the cosmos coherent and makes science possible. And yet not all coherence need be one of unequivocal necessity. My intuition is that more and more people are defining a
   base level or the initial state of a process in a way that goes beyond a characterization in terms of just local, empirical, observable properties to include real possibilities (some have spoken instead of
   real probabilities, but this raises a different issue) so that outcomes become unpredictable although probabilistically constrained by actual structures.

Well, sorry to run on like this, but you asked challenging questions that I couldn't just disregard. A lot of the points I've raised are discussed by C. Dyke, The Evolutinary Dynamics of Complex Systems:
A Study in Biosocial Complexity (NY: 1988). What you think of this book?
 
Haines Brown
... (expand)

2010-05-26
Can we make room for developmental constraints?
Reply to Haines Brown
Thanks very much for your 'reading list', Haines! I'll have a look into those papers as soon as I have time (which may, unfortunately, mean never -- I am not paid for studying functional explanation and the work I am paid for takes most of my time). 


In my view the notion of function with which Cummins (1972) is concerned is central to biology and the kind of explanation he calls 'functional analysis' or 'functional explanation' is one of the four main kinds of explanation in biology. It is however not the only kind of biological explanation that appeals to functions. In fact, most biologists would call Cummins-like explanations 'causal explanation' and use the term 'functional explanation' to refer to another kind of reasoning, one that is more like Hempel's 'functional analysis'. 

I coined the term 'design explanation' in 1995 to refer to the latter kind of reasonings after discovering that philosophers were confused by my use of the term 'functional explanation'. I sought for an alternative close to the language of working biologists, in a time when they frequently talked about ‘design considerations’ and ‘alternative designs' when discussing non-historical requirements on form, structure and organization. When I coined the term nobody in my country paid attention to the intelligent design movement in the USA, so I had no idea that the term 'design explanation' would be associated with intelligent design.

I call a property of a system organized if that property critically depends on the arrangement of the parts of that system and the timing of their activity in addition to the composition and the character of the parts (this notion of 'organization' is derived from Wimsatt's notion of 'emergence' in ‘Aggregativity: Reductive Heuristics for Finding Emergence’, Philosophy of Science, 64, pp. S372-S384, 1997). 'Design' is just another word for 'form of organization'. So my notion of design is independent of intention and does not imply a designer. It makes no assumptions at all about the way in which the organization is generated.

I am very sympathetic to the view that in order for an analysis/piece of reasoning to be called explanatory it must show that the phenomenon to be explained is somehow "necessary" and the article we are discussing in this thread (which is in part a reaction to Mahner and Bunge's "Function and Functionalism") spells out in which sense design explanations show that the phenomenon to be explained is necessary.



2010-05-26
Can we make room for developmental constraints?
Reply to Arno Wouters
Arno,

Thanks for the note. I don't want to drag things out, but I'm beginning to suspect there may be a certain convergence of views here.

I didn't realize that you were drawing on the Cummins article. He contrasts looking at things in terms of antecedent conditions rather than in terms of effects. I take this to be an implicit contrast between modal realism and empiricism, respectively. If I'm right, then the functionalism you are talking about seems a quite different kettle of fish than the usual. I'll have to pursue this, such as getting my hands on Predrag Šustar. (2007) "Neo-Functional Analysis: Phylogenetical Restrictions on Causal Role Functions." Philosophy of Science 74, 601-615. However, as far as Wimsatt's article is concerned, I paid it little heed, for it struck me as logical gymnastsics, appealing to a counterfactual argument to reconcile emergence and reductionism. I'm more sympathetic to physical rather than logical or linguistic argumentation.

I had assumed, and you didn't object, that living organisms are hard-wired to perpetuate themselves and so have a build-in Telos. This seems to imply living organisms have a physical measure of success. If they do not eat, they die, and so eating is a better course of action. If you happen to agree, it does have the interesting implication that all the talk about the need for consciousness to offer a measure of sucess is not accurate. However, this still leaves me with a difficulty. I'll try to convey it in terms less obscure than my usual.

I'm an organism. I assess my environment and see in it a food source and take action to seize upon it. My action seems "functional" in relation to my situation because the action is likely to support life. Had I instead tarried along the way so that the food source eluded my grasp, that course of action becomes less functional in relation to my needs. So it seems an explanation of the first successful action arises from the fact that I acted in such a way that it satisfies my needs. There seems an inclination in the literature to see this in terms of cause and effect; explained in mechanistic terms. Or can it?

Suppose the reason I tarried was to evade a predator, and as a result I took action that better preserved my life than getting the food right now. What I've done is to weigh courses of action in terms of which best ensures my life. I have filtered the causal relation between eating food and surviving to make it non-functional. So the functional relation of food and life, while it remains hypthetically true, is not actually operative. This functional relation of  food consumption and life is due to my filtering, to my constraint. Here functionality is constructed by my action and is not simply a relation of actualities. Let me put this another way. 

In mechanical causal explanation, it is assumed that the cause has certain empirical properties (a causal power that produces an effect that is a function of  its dispositions) as well as does the affected entity (its proclivity-the jargon usually used here slips my mind---to be able to change in certain ways). However, an organism seems to represent a non-mechanistic situation. It acts upon its environment in anticipation of a probable effect (survival); it is not the environment that acts upon it. So we have a need, which, although having a hard wired basis, is not a physical property but a pre-condition of life, standing in relation with a possibility in the environment, not actual empirical properties but a possible effect of properties.  I find it hard to reconcile this relation of an action to actualize possibility that will lead to a probable outcome with the old fashioned functionalism based on causality, a relation of actual empirical properties, rather than a relation of modalities.  The organism's action does not cause the food to sustain life, but its consumption constrains food's potential for actually doing so. I'm left with the feeling that even a probabilistic causality is inappropriate to this notion of functionalism and that a different notion of explanation is involved that does not look to necessity, but to possibility.

Haines


 



2010-07-14
Can we make room for developmental constraints?
Reply to Haines Brown
However, as far as Wimsatt's article is concerned, I paid it little heed, for it struck me as logical gymnastsics, appealing to a counterfactual argument to reconcile emergence and reductionism. I'm more sympathetic to physical rather than logical or linguistic argumentation.


Never think that Wimsatt is merely doing logical gymnastics! He is always trying to further science by providing conceptual tools to understand complex and messy situations. In the article to which I referred, he argues that a reductive explanation of the behavior of a system to the behavior of the parts of that system does not license the conclusion that the system's behavior or the system itself is 'nothing but' the sum of the behavior of the parts. This is because reductive explanations rely on well-chosen decompositions (of the system into parts) and assumptions about the organization remaining intact. You might call that an attempt to reconcile reduction and emergence if you like, but, imho, that does wrong to the richness of the article. 

2010-07-14
Can we make room for developmental constraints?
Reply to Haines Brown
Haines, I do not understand everything you say, but I agree with most of what I understand and I do sense a convergence of ideas here, although I am not sure how exactly,

You write:

I had assumed, and you didn't object, that living organisms are hard-wired to perpetuate themselves and so have a build-in Telos. This seems to imply living organisms have a physical measure of success. If they do not eat, they die, and so eating is a better course of action.


I agree (more or less - I do not think that from 'if they do not eat, they die' it follows that eating is always the better course of action).

 
If you happen to agree, it does have the interesting implication that all the talk about the need for consciousness to offer a measure of sucess is not accurate.


*All* the talk about the need for consciousness to offer a measure of success? I am afraid I get caught now, but I don't know about any such talk.

Concerning your example of an organism that tarried to evade a predator and missed the opportunity to seize upon a food source you write:

So we have a need, which, although having a hard wired basis, is not a physical property but a pre-condition of life, standing in relation with a possibility in the environment, not actual empirical properties but a possible effect of properties.


I have the impression that you are confusing different notions of needs and different kinds of explanation.  As I explained in my "The Functional Perspective of Organismal Biology" (2005), I think (as do many biologists)  that in order to understand living systems one needs a combination of explanations of four kinds to explain organisms and their behavior: mechanistic explanations, design explanations (usually called functional explanations), developmental explanations and evolutionary explanations.

As you indicate, in your example the mechanistic explanation might be that he organism evaluated its actions according to its present needs and concluded that it was best to give priority to avoiding predators over seizing a possible source of food. This means that the organism must have a hard wired representation of what is worth getting, as well as a mechanism to compute the score of different actions on these values and, perhaps a mechanism to change the computation based on the results of the chosen action. This explanation talks about needs as that what the organism values. Let's call needs in this sense subjective. These needs are hard-wired into the organism.

There is another sense of 'need', namely: that what the organism actually needs to stay alive. Let's call this the objective needs. Objective needs are hard (if the organism does not satisfies them it dies) but unlike subjective needs not hard-wired. As you aptly say, objective needs are pre-conditions for life. But these are not the kind of needs that figure in the mechanistic explanation.

I agree that the mechanistic explanation above does not suffice to understand the behavior, In my view, the very nature of a mechanism as a system *organized* for the production of a certain behavior means that a mechanistic explanation never suffices to understand a mechanism. In order to understand the behavior of an organism the mechanistic explanation must be supplemented with a design explanation, a developmental explanation and an evolutionary explanation.

I don't see that the mechanistic explanation above presents the behavior of the organism as the passive result of the environment acting upon the organism. In my view, a choice mechanism is a paradigmatic example of an active response to what happens in the environment,

The design explanation will start with the 'observation' that it is an objective need that the subjective needs are such that they help the organism to satisfy its objective needs and will continue by showing that the subjective needs satisfy this constraint but it cannot explain how that state (the state of the subjective needs being such that they help the organism to meet its objective needs) came about. For that reason it is not a causal explanation.

Developmental and evolutionary explanations explain how the mechanism came to be the way it is and hence how come that the subjective needs are tailored to the objective needs,

I am afraid that I get caught a second time, for I have no idea what you mean when you talk about "the old fashioned functionalism based on causality". 

I'm left with the feeling that [...] a different notion of explanation is involved that does not look to necessity, but to possibility


In my view, design explanations are concerned with what mechanisms can be alive, so in that sense they are concerned with possibility rather than necessity. However, because it is necessary that the actual form is one of limited number of possible forms, I am of opinion that design explanation show that the phenomenon to be explained is somehow necessary. (Newton's laws do not show that the moon necessarily circles around the earth, only that this motion is one of the possible forms in which a two-body system can exist!)


2010-07-14
Can we make room for developmental constraints?
Reply to Haines Brown
PS: Haines, I would appreciated it if you fill in your profile on philpapers with a bit of background ('not much to say" says not much ;-) ), an e-mailaddress and a list of your papers! It is nice to know to whom I am talking!

2010-07-15
Can we make room for developmental constraints?
Reply to Arno Wouters
A range of issues have come up, and I'll do my best to stick to basics. Not always easy for me.

You said, "...further science by providing conceptual tools to understand..."  Without arguing the
case, it seems to me there's a tradition in the modern western philosophy of science to privilege
epistemology over ontology. This means in very simple terms that the trend is to draw inferences
from observations. Again, in crude terms, this means that the only aspect of the world that can
serve as the basis of knowledge is that which impinges on the senses or on instrumentation. I
won't argue the case here, but it is easy to show that this privileging of epistemology is clearly
a reflection of particular historical circumstance, that there is an alternative way to engage the
world that conveys a much richer notion of the world than possible in observation, and that since
we now appreciate the extent to which observations themselves are mediated and not simple reflections of the world, a critical reconstruction is needed to avoid the radical skepticism that
has prevailed in the social science in the past two decades.

I brought up "hard-wired to perpetuate themselves," not because it raises the issue of the
measure of success, but in relation to ontogeny (off hand this seems the old debate over
functionalism in different form). At at certain early point in evolution from complex
hydrocarbons to "life", one feature is a mechanism of self-perpetuating structure. My
impression is that in the lab one can build a complex molecule to do this to some extent,
but my ignorance here is profound, and I can only guess that this behavior was accidental,
and once it existed it naturally became generalized. For example, generally speaking,
living organisms are far-from-equilibrium systems that must continually dissipate an energy
gradient to maintain their structure. Arguably, these "structurally emergent" systems have
a greater tendency of generating improbable outcomes, so there may be an intimate  tie
between being a far-from-equilibrium system (not uncommon in nature), having a physical
boundary, and self-perpetuation. But, again, I'm ignorant about the empirical specifics that
would lend weight to such a hypothesis. Note that my mechanistic description of the
implied feedback loop does not mean the levels are not emergent, and arguably feeback
(in contrast with mere causal effect) depends on the levels involved being emergent (I skip
over the argument in support of this generalization).

The talk about "consciousness" as necessary for a measure of success, not surprisingly,
is mostly found in the social sciences (the old debate over historical progress, for example).
In biology I assume the organism modifies its properties or behaviors in such a way as
to ensure the energy gradient needed to maintain the system in its far-from-equilibrium
state. It seems that this constitutes "needs" of the system becayse in their absence a
significant property or behavior of the whole would cease to exist (in human affairs
I'd argue that needs are necessary emergent, but that would be another issue). In bio
systems, a need not met ultimately means a dissipation of the system.

Now you seem to raise an objection to this, but I don't see that I'm mixing explanation
and needs. A tropical storm is a far-from-equilibrium system, but has no "need" to move
contrary to the Coriolis effect in order to preserve itself. But living organisms can do that,
such as locating and ingesting food. This seems entirely independent of consciousness.
On the other hand, explanation is a mental operation that satisfies the mind. Traditionally,
this was the discovery of necessary relations. Today, things seem to have become more
complicated (singular causality and emergence in the historical sciences like bio-
evolution, for example).  I'd boldly suggest that to date there exists no explanation of
emergent systems.

I'm rather dismayed by your four types of explanation for a host of reasons. I can't go
into this in detail, but for example, I see mechanistic explanation as a hypothetical
limiting case and even then not entirely accurate even for systems that are physically
or conceptually closed. Functional explanations, at least in social science, have been
roundly criticized for being circular or teleological (begs questions). "Developmental"
sounds to me like emergent processes, and in social science (particularly for the
mind-body problem), there is a widespread intuition today that the base level needs to
be defined to include extrinsic or exogenous unobservables such as possiblities, and I
suspect (because of ignorance), that this is true of bio-evolution as well. That is, my
instinct is that your four categories come down to the same thing: a structure that
constrains possibilities and a relation of structures that actualizes (localizes) these
possibilities, with the probability of the outcome a function of the degree of closure.. 

You also loose me with your distinction of subjective and objective needs, I suppose in
part because I don't accept the categorical contradiction between the two. All action
involves feedback loops between synapses and sensory apparatus. Now, mechanistic
explanation is reductionist in that behavior is unequivocally determined by sensory inputs,
and it seems you object that this is too simple a description of animal behavior, which is
to some extent always a bit unpredictable or complex. So far I'm with you, but from this
point, I'm unclear. I assume we are speaking of emergent levels, not mechanistic feedback
loops. There seems to have been a lot of criticism of the standard neo-Darwinian model.
Rather than mechanistic causality, P. Y. Quenette and J. F. Gerard, "Why Biologists Do
Not Think like Newtonian Physicists," Oikos, 68 (1993), 361--363, suggest the primacy
of constraint on real possibilities, which are systematically excluded in a mechanistic
picture of the world.

There is no empirical correspondence between stimulae and reactions to them, but we
can say that the latter suit the former, which you seem to refer to as a functional relation.
But here you  loose me. Such an association of properties of levels seems a Humean
constant conjuncture and thus descriptive rather than explanatory. A dissociation of
that functional relation leads to predictable outcomes (such as death), but prediction is
not explanation, but merely reason to have confidence in one's predictions. Often in the
natural sciences, one can successfully employ a reductionist methodology to analyze
each link in a causal chain, but then this is no longer a functional explanation, but a
mechanistic explanation. And your word "design" surely implies more than just functionality,
and seems to imply an ideal model. If I cross the stream by hoping from rock to rock,
each rock functions as a stepping stone, but the ideal model is limited to my conscious
mind when it infers from seeing the rocks that I might cross the stream by using them.

We seem to agree on possibility, but you loose me when you associate possibility with
being alive. Maybe the problem is that I assume a modal realism. In predicting the weather,
the coming storm might hit my city or it might not. The weatherman describes the situation
in terms of a probability distribution. My assumption is that such a distribution is real and
need not have anything to do with living systems or consciousness. Its probability of hitting
my city is quite independent of the weatherman. Now, when a possibility is actualized, I
don't see why you speak of it as necessary. The storm front always has a probability
distribution for its future trajectory. What actualization means is only spatio-temporal
localization, and I'd ague that localization arises from the structure of one process entering
into a physical relation with the probability distribution of another. For example, for the
storm to have hit my city means I impose a geographic frame of reference to measure
the storm's proximity. Without a frame, the location of the storm is an unactualized
possibility. All measurements in science are a relation of systems that constitute a
frame of reference that specifies the effect of one system on another, such as in
observation. Hence it is objective in that the probability distribution belongs to the object
of study, and subjective in that the actualization of this probability distribution depends
on the subject.

Haines Brown





2010-07-16
Can we make room for developmental constraints?
Reply to Haines Brown
Haines, I do not want to suggest that you confuse needs and explanation. My impression was that you confuse objective needs and subjective needs and also that you confuse design explanations and mechanistic explanations that appeal to needs and/or functions.

In my view mechanistic explanations can appeal to subjective needs, and your explanation of an organism's tarrying because it judged that evading a predator was a better course of action than looking for food is a case in point. 

Similarly, mechanistic explanations in biology typically appeal to causal role functions (functions in Cummins's sense) but that does not make them functional in the sense biologists use that term. Biologists call mechanistic explanations (including Cummins-explanations) 'causal explanations' even if these explanations appeal to role functions. This is one of the reasons why the term 'functional explanation' is confusing and why I coined the term 'design explanation' to refer to  the explanations that biologists call 'functional explanations'.

What distinguishes design explanations (the kind of explanations that biologists call 'functional explanations') from mechanistic explanations (the kind of explanations that biologists call 'causal explanations') is neither the appeal to needs as such, nor the appeal to role functions but the appeal to what I, in my previous post, called objective needs, that is appeals to what Cuvier called 'the conditions of existence' and what you seem to call the pre-conditions.

In my view the failure to distinguish objective needs from subjective needs and the failure to see that design explanations are not causal easily slips into unacceptable forms of teleology. This occurs for example when one would say that an organism ate some food because it needed it to stay alive. If 'need' is read as 'subjective need' the claim that some organism ate something because it felt the need to do so can be part of a mechanistic explanation that articulates part of the causal network that brought about the eating. If 'need' is read as 'objective need' the claim that certain types of organisms eat because they need to stay alive might be part of a design explanation that articulates part of the network of functional dependencies that make it possible to stay alive. However if one reads 'need' as 'objective need' and at the same time sees the claim as part of an explanation that brought about the behavior, one ends up in an unacceptable form of teleology in which objective needs bring about the behavior that satisfies them. 

By the way: this is exactly why functional explanations in biology would be unacceptable teleological if the accounts of functional explanation by Charles Taylor (*The explanation of behavior*, 1964) and Larry Wright (*Teleological Explanation* 1976) were accurate. Fortunately ;-) these philosophers have no idea how functional explanation in biology works.

So in my view design explanations are not teleological exactly because they are not of a causal/mechanistic nature. But this also means that design explanations by their very nature cannot explain how the behavior explained by the design explanation came about. For that reason design explanation should be supplemented with causal explanations in order to understand the organism. 

At the same time, causal explanations do not suffice to understand the organism because these explanations only show how the problem to stay alive is solved, but we also want to know what the problems of being alive are, why a certain problem is a problem and why one design (form of organization) can solve the problem and another not. So we need design explanations in addition to mechanistic explanations in order to understand organisms (even stronger: we need design explanations to understand any mechanism).

In their 'Function and Functionalism' (2001) Mahner and Bunge (to which you refer in one of your post in this thread) are puzzled that I talk of design *explanation* although I argue extensively that these 'explanations' are not causal. In their view explanations are by definition causal and the reasonings that I call 'design explanations' are according to them 'descriptions or perhaps subsumptions but not explanations'.

In the article we are discussing in this thread I answer that if explanation is by definition causal, the kinds of reasoning that I have called 'design explanation' are not explanations indeed. However in my view the difference between descriptions and explanations is not that the latter are causal and the first not. In my view descriptions show us what is the case and explanations show us how what is the case hangs together. Causal influence is one form of hanging together, functional dependence is another one. I am open to other relations by which things can hang together (I suspect that 'normatively justified by' is a way in which behaviors in the social world hang together in addition to causal influence and functional dependence). Mere correlation is in my opinion not a form of hanging together and this is because hanging together implies a form of necessity, albeit a very weak one and the article spells out this weak form of necessity (as did my dissertation).

Another way of distinguishing descriptions from explanations is this: descriptions describe what is the case, explanations relate what is the case to what is possible. I am sympathetic to this view too and design explanations do relate what is the case to what is possible, as does for instance the Newtonian explanation of why the moon circles around the earth (I do not associate possibility with life, as this reference should make clear). However,  I am not sure that causal explanations relate what is the case to what is possible and as I do feel that the kinds of reasoning normally called 'causal explanations' are explanatory I prefer the view that explanations differ from descriptions in that they show how things hang together, over the view that explanations differ from descriptions in that they relate what is the case to what is possible.



2010-07-16
Can we make room for developmental constraints?
Reply to Haines Brown

Now, mechanistic explanation is reductionist in that behavior is unequivocally determined by sensory inputs, and it seems you object that this is too simple a description of animal behavior, which is 
to some extent always a bit unpredictable or complex.


I do think that the picture of animal behavior as unequivocally determined by sensory inputs is very wrong, but that was not my objection.

My point was rather that I don't think that mechanistic explanations by their nature picture the behavior of an organism as determined by the sensory input. In my view mechanistic explanations have all the sources to take wiring and inner states into account. The explanation of an organism's tarrying because it evaluated possible actions according to its present needs and concluded that it was best to give priority to avoiding predators over seizing a possible source of food is a perfect example of a mechanistic explanation that does not view the organism's behavior as determined by the sensory input. It views it as determined by a combination of its sensory input, its wiring, and its inner states.

A related point was that I don't see that such a mechanistic explanation pictures the organism as acted upon by the environment rather than as acting upon the environment in anticipation of a possible effect. According to the mechanistic explanation you suggested the animal calculates the score of the effects of possible actions on its hard-wired values. I would call that acting upon the environment in anticipation of a possible effect.

So if reductionism is defined as an attempt to explain behavior in terms of sensory input, mechanistic explanations are not reductionist.

2010-07-16
Can we make room for developmental constraints?
Reply to Haines Brown
Haines writes:


Often in the natural sciences, one can successfully employ a reductionist methodology to analyze each link in a causal chain, but then this is no longer a functional explanation, but a
mechanistic explanation.




Haines, I get the impression that we are talking at cross purposes. In the first sentence of this quote you suggest that if one employs a reductionist methodology to analyze a causal chain a functional explanation is changed into a mechanistic one. I feel very stupid for I have no idea what you mean.

Suppose you and I are studying a biochemical pathway of a certain type of organism (let's call organisms of that type t-organisms). The pathway we are studying converts molecule A into molecule B and molecule B into molecule C.

We might start with a design explanation. For example we might be able to establish that C is needed for t-organisms to be able to move, and that C is not present in the environment in which t-organisms live, but A is and A can be easily converted into C via A.


Now we analyze each link. We might be able to get a picture of all the enzymes involved in both steps and how the enzymes attach to A, B, and C, how the configuration changes when A,B, and C bind to those enzymes, and we may employ thermodynamics to explain how this leads to changing A into B and B into C. Than we have a mechanistic explanation. You might call this a reductionist strategy if you like, although I want to emphasize that this explanation heavily relies on viewing the parts in the context of the whole.


I don't see that the design explanation changes into a mechanistic one as the result of  the application of the reductive strategy, which is what you seem to suggest. In my view we have two explanations: a design explanation that explains why it is useful for t-organism to convert A into C and a mechanistic explanation that explains how they do so. Both explanations are needed to understand the pathway and so they are complementary.



And your word "design" surely implies more than just functionality, and seems to imply an ideal model.


Please explain what the ideal model in the above sketch of a design explanation for a biochemical pathway,



If I cross the stream by hoping from rock to rock, each rock functions as a stepping stone, but the ideal model is limited to my conscious mind when it infers from seeing the rocks that I might cross the stream by using them.



I have to apologize again, for I have no idea what the point is of your remark.

Suppose you are hiking in the mountains and there is an unexpected river across your path. You look at the river, notice rocks in it, picture yourself using the rocks as stepping stones and conclude that you might be able to cross the river by using the rocks as stepping stones. You do so and succeed.

It is easy to imagine a design explanation of why you needed to use the rocks as stepping stones: it is one possible way (perhaps the only one) in which organisms that walk on two legs of the length you have can cross a certain river: the stones were near enough to each other, there was not so much water in the river that you would be swayed away when you stepped on them and so on. This design explanation can be elaborated with explanations why other forms of crossing would not be possible. This design explanation does not appeal to an ideal model, but only to what is physical possible and what not.

The mechanistic explanation of how you managed to cross the river on the other hand does appeal to an ideal model, namely you picturing yourself as using the stones,

Of course, the fact that some mechanistic explanations appeal to ideal models does not mean that mechanistic explanations in general imply an ideal model. Neither does the fact that some design explanations appeal to ideal models means that design explanations require appeal to ideal models. 

2010-07-16
Can we make room for developmental constraints?
Reply to Arno Wouters
Haines, I do not want to suggest that you confuse needs and explanation. My impression was that you confuse objective needs and subjective needs and also that you confuse design explanations and mechanistic explanations that appeal to needs and/or functions.
If I understand correctly, you agree that identifying a system need does not represent an
explanation, for a need is a characteristic of the system itself, not its necessary relation
with the world. How it arrived at that state in relation to its world requires a different kind
of explanation. Is this inference correct?

But the issue is over subjective and objective and my confusion of them. I plead guilty and
worse, deny the distinction! I believe/hope I can explicate the point without having to rely on
the more challenging thrust of your message re. function and design vs. mechanistic
explanation. That is, I will discuss it one-sidedly in mechanistic terms.

A meteorite hits the earth, shatters and leaves a small crater. The meteorite produces an
effect on the earth and the earth has produced an effect on the meteorite. It makes no sense
here to speak of subject or object, or for that matter, perhaps cause and effect, for both the
earth and the meteorite are both. Now I consider a biological organism, say a clam, and I
find that when touched, it closes its shell. The mechanical stimulus of my touch sends an
electro-chemical signal to its ganglion, which is hard-wired to processes such signals and
send an appropriate electro-chemical signal back to the muscle to close the shell.

Now I will be picky. This scenario actually describes something happening in my mind,
which, to avoid contradiction, invents a time line on which events are placed (McTaggert
 B-Time) or have an extrinsic temporal property (McTaggert A-Time). But the truthmaker
objection points out that past events can't influence present events, for the situation is
more like a Markov process in which an actual situation defines the probability distribution
of possibilities. Time is not relevant. That is, the act of my reaching out to touch the clam
and the clam's sensory nerve ending sending a signal are both cause and effect. True, the
effect on me is small, but it is real: my haptic sense of touch sends a signal to my brain.
It represents a causal interaction rather than my being simply agent and the clam the
object of action (actually, I am inclined to conflate cause and effect and instead speak
of a processual superposition, but that's another story). This kind of analysis applies to
the entire "causal" chain in the above scenario. The point here is, what is subject and
what is object? The distinction is lost, at least in ontological terms. When I touch the
clam, as far as my haptic apparatus is concerned, I am object and the clam is subject,
but when we trace the signal to the clam's ganglion, etc., the clam become object and
I am subject. Better, the whole represents one process in which my action and the
clam's action combine.  

I have analyzed this scenario in mechanistic terms, but in fact, an electro-chemical
signal traveling to the clam's ganglion is empirically unlike the mechanical motion
of my touch. I would argue that this signal is an emergent effect, although here I
define "emergence" in ontological terms rather than the conventional epistemological
terms in which predictability is definitive. For me, predictability is an accidental effect
of relative closure. My description of the scenario is conceptually closed, and therefore
my analysis of it in mechanistic terms is approximately true. 

The distinction of subject and object might make more sense if action is the result of
a mental event such as intention. But I'd extend the analysis even to this case. However,
the hypothetical limiting case of closure seems inappropriate here, for there is a
broad probability distribution, so that the constraint of my intention has an outcome
even less predictable in relation to the initial state of the system or external dertermina-
tions. That I am "subject" and the clam the "object", is only an artifact of the limitations
of mind. 

I suspect you are unsympathetic to the above argument, but I had to drag it in to be
able to address your principle concern for function and design.

You start by distinguishing two senses of "function", that of Cummins and that used
by biologists. The issue here seems to be whether a "role function" is causal. The
problem is that (if you will pardon my putting it rather strongly), no one has the
slightest idea what "causality" is! There is a variety of hypotheses, no one of which
has stood up. I won't belabor this point here, but at least it seems that, given the
uncertainty, one must start with a cogent definition of causation.  For example, very
popular is a variety of transference theories, in which case one must ask what is
physically transferred between a role function and the system in which that role is
carried out? I suspect the answer would rely on some kind of counterfactual argu-
mentation, which might make sense in philosophical terms, but probably not physical. 

You proceed to distinguish function, which we may agree is descriptive rather than
explanatory, from objective needs or my preconditions of existence. Intuitively these
seem real, although not causal. Perhaps, to be physical rather than philosophical,
they can be seen as boundary conditions? If so, then they would define degrees of
freedom, or in my preferred terminology, the probability distribution (that is, the
state of being alive requires that a mechanism of metabolism be highly probable.)

If I understand correctly, you are objecting to the approach that would translates objective
facts into causes of subjective behavior. Surely you are correct, and I believe you here
touch upon the reason why Darwinian explanation has found so much appeal in the
social sciences. I get the feeling this appeal is because objective factors are not causes,
but represent possibilities that are "seized upon" by the organism, which interacts with
its environment. You point out that we need design explanation in any mechanistic
explanation to account for results that are emergent.

Where we seem to part ways is that, if the above summation carries any weight, I take
your point to a far more radical extreme: I apply it to all domains, mechanistic as well
as human, for I replace causal explanation with what I call processual superposition.
I believe this helps me address some of the "Irish pennants" left over from your
exposition. However, I'm unacquainted with the extensive literature on teleological
explanation in biology, so please forgive my ignorance.

One such hanging pennant is the issue of teleology. Roughly, it seems to me that an
unacceptable teleology is one in which ideal factors determine a trajectory, while a
legitimate teleology is on in which physical factors (including unobservable) determine
a trajectory. However, to me, a "trajectory" is a mental construct not found in nature (I'm
inclined to deconstruct time), for I reduce the future to the probability distribution of the
"present". I see a probability distribution as arising from the constraint of structure on
possibilities, and it makes no difference if this structure is external (the structure of the
causal process), internal (the structure of the process of interest), or mental (the
structure of the mental process such as intention). (You might correctly infer that I
perversely discount the conventional distinction of external/internal, inside/outside). 

You make the conventional distinction between genetic or historicist explanation and
structural explanation. It should be clear from my ramblings that I doubt there is such
a contradiction between abstract and concrete particularity. A fondness for genetic
explanation is now seen outside biology, such as the trend toward singular explanation
in the natural sciences and narrative explanation in historiography. But it seems to me
that the contradiction between abstract structure (a real structure that emerges from
the relation of constituents) and concrete particulars is a peculiarly modern Western
perspective that has been shown to be inadequate. An emergent structure, in my view,
represents the constraint of the structure of one constituent on the real possibilities of
another, and their relation actualizes the resulting probability distribution as an
observable systemic effect.   

Another pennant is whether explanation is necessarily causal. It certainly was in
terms of positivism, but I'm not so sure it is universally true otherwise. The ideological
ground of modern science is a presumed cosmic coherence (16th century neo-platonism).
That is, things have a "necessary" relation, but  this word does not mean a mechanistic
or causal relation: it means one thing is indispensable to another, which it cannot do
without. My (as you now know, weird) perspective is that this necessary relation is
one of enabling (offering constrained possibilities) rather than a causal determination.
If so, this would change the nature of explanation, for you don't look for the cause of
an observed outcome and validate it by successful prediction, but instead seek to
ascertain its possibilities and the inherited constraints that define the probability of
those possibilities. The actual empriical outcome is therefore is probabilistic rather
than determinant.

I like to think of this as the essence of  Darwinian explanation, but I suspect you will
strongly object. Nevertheless you say, "descriptions describe what is the case,
explanations relate what is the case to what is possible," so maybe we are traveling
the same road after all. Your not associating possibility as peculiar to life strikes
me as obvious. It simply presumes a modal realism (not Lewis' popular plurality of
worlds view). However, rather than conventional modal realism, I anchor the modalities
of potency and possibility on structure (vs. Aristotle's and Lewis' separation).

I'm sure much of what I've been saying must strike you as hopelessly obscure or
outright wrong, but my intuition is that on closer inspection we are saying much the
same thing.

Haines