Philip Kitcher showed a way to save the concept of progress and truth in science in his latest book, The Adv
ancement of Science (1993, hereafter, AS). This book contains many
good arguments to defend objectivity in science, but in this paper I try
to show that his argument ends up with almost the same anti-realistic view as in Larry Laudan's Progress and Its Problems (1977, hereafter, PIP). Kitcher tried to differentiate his theory from Laudan's, but his attempt seems to fail in some important points. For this purpose, first I will clarify Kitcher's and Laudan's arguments about progress and ration
ality. Next, I compare these theories and point out where is the crucial difference between them. The result of this comparison will require Kitcher to give further argument how his theory can be different from Laudan's. I don't think Kitcher is very suc
cessful in this task.
Start with summarising Laudan's argument.
In Laudan's view, the fundamental aim of science is problem solving. Scientific problems can be classified into empirical ones and concep
tual ones. Empirical problems are also classified into three kinds: solved problems, unsolved problems, anomalous problems. An empirical problem is solved "when, within a particular context of inquiry, scientists properly no longer regard it as an un
answered question" (PIP 22, hereafter the pages are in PIP unless otherwise noted). Usually unsolved problems are ignored or unknown (because if it is not ignored, it is called an "anomalous problem"; this is only a matter of terminology).
When a theory answers an unsolved problem, this problem becomes an anomalous problem to other theories (29). Laudan's concept of scientific progress is explained in terms of these problems; "one of the hallmarks of scientific progress is the transfo
rmation of anomalous and unsolved empirical problems into solved ones" (18). But for a full account of progress, we need to add conceptual problems. Conceptual problems are "higher order questions" (48), which are deduced from empirical con
tents of the theory. There are two kinds of conceptual problems. Internal conceptual problems occur "when T exhibits certain internal inconsistencies, or when its basic categories of analysis are vague and unclear"(49). External conceptual probl
ems occur "when T is in conflict with another theory or doctrine T', which proponents of T believe to be rationally well founded" (49). In terms of these problems, Laudan define a notion "problem-solving effectiveness"; "the ove
rall problem-solving effectiveness of a theory is determined by assessing the number and importance of the empirical problems which the theory solves and deducing therefrom the number and importance of the anomalies and conceptual problems which the theor
y generates" (68). This notion is directly related to the definition of progress; "progress can occur if and only if the succession of scientific theories in any domain shows an increasing degree of problem-solving effectiveness"(68).
N
ext Laudan turns to the problem how to decide what counts as a problem, and what counts as a solution. The concept of a "research tradition" is introduced to answer these questions. A research tradition is "a set of ontological and methodol
ogical 'do's' and 'don'ts'"(80). The ontology tells us what is a significant problem and what is not. methodology tells us what is the right way to solve the problems. A research tradition can change by modification of its subordinate theories or by
changing some of its core elements (96). This latter does not necessary mean a change to another research tradition. A research tradition can evolve with minor changes. In Laudan's theory a scientific revolution doesn't play an important role in theory ch
ange (133-138).
The evaluation of research tradition depends on the evaluation of the theories in the tradition (106-114). The adequacy of a tradition is judged by the problem solving effectiveness of the latest theory in the tradition. The progressi
veness of a tradition is judged by general progress and the rate of progress. The general progress is measured by comparing the adequacy of the oldest theory in the tradition and that of the latest one. The rate of progress is measured by the change of ad
equacy in an identified time span. With these standards, we can decide which theory we should accept, and which theory we should pursue. In the context of acceptance, we should "choose the theory (or research tradition) with the highest problem-solv
ing adequacy" (109). In the context of pursuit, "it is always rational to pursue any research tradition which has a higher rate of progress than its rivals" (111).
In his account, the rationality in science is derived from progress. He
says, "the single most general cognitive aim of science is problem solving" (124). And rationality is defined in terms of problem solving and progress; "[t]o make rational choices is , on this view, to make choices which are progressive (i
. e. which increase the problem-solving effectiveness of the theories we accept)" (125). Rationality is not relative to a research tradition. Problems and the solutions are relative to a research tradition, but he thinks he can show the way to compar
e different theories and different traditions. Even if our words to describe the problems are theory-laden, if they can be translated into the words in other theories, we can compare the problem solving effectiveness (142). Even if we don't have such a tr
anslation, we can compare the progressiveness of the theories (146). He admits also non-epistemic elements can affect scientific theories, and this effect can be rational (132). In these points, his rationality is not relativistic.
But his rationalit
y is not absolute either. Progressiveness can be measured only when we compare two theories or traditions (120). In addition, his rationality has nothing to do with truth; "I don't even believe, let alone seek to prove, that problem-solving ability
has any direct connection with truth or probabilities"(123).
More importantly, progress is not cumulative for him (147-150). This is mainly because different traditions have different way of weighting problems. When a new theory solves a new pro
blem and cannot solve several old ones, we may think the new solution is significant enough to ignore the loss of old problems.
It seems to me that this last points run counter to his assertion of comparability of traditions (for example if a traditi
on thinks much of trivial problems and thus can produce many solutions to their problems, we should choose this theory instead of other theories which grapple with difficult problems; this is counter intuitive). This may be interesting problem, but I don'
t want to argue it in this paper.
Next we should turn to Kitcher's argument. There are several similarities and strong contrasts between Kitcher and Laudan. This is not in any way accidental, because Kitcher himself repeatedly compared his theory wit
h Laudan's.
Kitcher on progress
In Kitcher's account of science, scientific practice has several components (AS 74, hereafter the pages are in AS unless otherwise noted); the language, the significant
questions, the accepted statements, the set of patterns to answer questions, the standard examples of credible informants, the paradigm of experimentation and observation, exemplars of good and faulty scientific reasoning. Among these components, the most
important part of the practice is patterns to answer significant questions, namely, explanatory schemata (82-84). Explanatory schemata show us the schematic way to give explanation to a significant question, and how to fill in the schema by replacing the
dummy letters in the schema.
These many components in scientific practice help him to introduce the concept of "broadly progressive" (92). Kitcher thinks if science is progressive, there should be "general upward trend" (90). T
his does not necessarily mean a successor practice should always be progressive in all dimensions. In many cases, successor practice is progressive with respect to its predecessor in several components and not progressive in other components. But when we
compare the oldest practice and the latest one in the sequence, if we find the latest one is progressive in all components with respect to the oldest one, the sequence has the general trend to progress. this is called "broadly progressive", and
Kitcher thinks science is progressive in this sense.
Kitcher points out two important kinds of progress, namely "conceptual progress" and "explanatory progress" (95). Conceptual progress is a progress in language. If a successor pr
actice improves the language in the way the words refer to naturalþ kinds more accurately or refer to something we can specify, it is conceptually progressive (104-105). Explanatory progress is a progress in explanatory schemata. If a successor practice
has more correct schemata or improves a schema over a predecessor one, it is explanatory progressive (111). What does correctness of a schema mean? "[A] schema is correct if it identifies a class of dependent phenomena and specifies some of the entit
ies and properties on which those phenomena depend" (111). A question is significant when it corresponds to a explanatory schema, and genuinely significant when it corresponds to a correct explanatory schema. A statement is significant if it is a pot
ential answer to a significant question. These significant statements become significant truths by way of improving language. After all, his two kinds of progress aim at significant truth.
This notion of truth obviously presupposes a realistic worldv
iew. He gives several arguments to defend realism. It is not the purpose of this paper to examine his defense on realism, but we should take a look his argument on the relationship between progress and realism. He puts forward a question, "[h]ow can
we know that we are making progress in any of the senses that I have delineated?" (128). His answer goes as follows; "[w]e have scientific views about the relations between ourselves and the rest of nature" (134). Thus we can investigate h
ow our knowledge are correspond to the nature. But as he admits, this argument is circular (135). He exclude from his argument those who doubt these "scientific views" themselves, but he doesn't give any convincing reason to do so. I come back t
his point later.
Kitcher on rationality
Now we know what is the progress in his view. He proceeds from this concept to rationality. When are we rational in science? He derives the concept of rationality from the concept of progress. The aim of sc
ience is to progress, and rationality is "a means-end notion" (179) in this context. So we are rational when we are doing something which contributes to progress.
First he reconstructs the notion of rationality in the way suitable to his pic
ture. His standard for rationality is called the "external standard (ES)" (189).
(ES) The shift from one individual practice to another was rational if and only if the process through which the shift was made has a success ratio at least as high as that of any other process used by human beings (past, present, and future) across the set of epistemic contexts that include all possible combinations of possible initial practices (for human beings) and possible stimuli (given the world as it is and the characteristics of the human recipient).
The point of this standard is that the process of shift should have a success ratio at least as high as that of any process used by human beings" (189). He relaxes this condition soon. In these rel
axed standards, for example we don't have to know future success of the process, and also we don't have to maximize the success ratio. By this change many things can be counted as rational. So he suggests that we "dissolve" the notion of rationa
lity itself into "goodness in cognitive design" (194). But the core of the standard remains the same; rationality is judged by the success ratio in progress.
He uses (ES) to construct a model of rational consensus. His "compromise model
" (201) is an attempt to derive a rational decision-making from anti-rationalistic initial conditions.
Compromise model
(C1) The community decision is reached when sufficiently many sufficiently powerful subgroups within the community have
arrived at decisions (possible independent, possible coordinate) to modify their practices in a particular way.
(C2) Scientists are typically moved by nonepistemic as well as epistemic goals.
(C3) There is significant cognitive variation within sc
ientific communities, in terms of individual practices, underlying propensities, and exposure to stimuli.
(C4) During early phases of scientific debate, the processes undergone by the ultimate victors are (usually) no more well designed for promoting
cognitive progress than those undergone by the ultimate loser.
(C5) Scientific debates are closed when, as a result of conversations among peers and encounters with nature that are particularly produced by early discussion to modify individual practi
ces, there emerges in the community a widely available argument, encapsulating a process for modifying practice which, when judged by (ES) (and other standards canvassed in earlier sections), is markedly superior in promoting cognitive progress than other
processes undergone by protagonists in the debate; power accrues to the victorious group principally in virtue of the integration of this process into the thinking of members of the community and recognition of its virtues.
(C1)-(C4) are something an ti-rationalists also can admit (with small changes in terminology). (C5) is a very rationalistic criterion. A natural question here is whether it is possible that people in these conditions can achieve rational consensus. This is also an important questi on for my enterprise in this paper. I argue it later.
Now we summarize Kitcher's theory. For Kitcher the unit of progress is a scientific practice. Progress means to proceed toward significant truth. Rationality is defined in terms of progress.
Comparison between Kitcher and Laudan
Kitcher's argument has a similar structure as Laudan's. Both of them claim we can restore the rationality of science, and they agree rationality can be defined in terms of progress. Their view
s on progress also have some common characteristics. Kitcher's "scientific practice" includes some items which Laudan's "research tradition" doesn't have, but these items don't play important role in Kitcher's argument. Laudan's "
problem-solving" can be compared with Kitcher's "explanatory schemata". In Laudan's view, the progress of a theory in empirical problems includes finding new significant problems and solve them. This is almost what Kitcher means by "ex
planatory progress". Kitcher's "conceptual progress" is expressed as evolution of research tradition in Laudan. For conceptual progress means refinement of language about natural kinds, while Laudan's research tradition contains "ontol
ogical and methodological do's and don'ts", and the evolution of research tradition includes refinement of these ontologies. Laudan's "conceptual problems" doesn't have exact counterpart in Kitcher's theory. In some cases the solution of co
nceptual problem may be the same as "conceptual progress". In other cases it can be interpreted as tension between the components of practice (86n). Anyway, in many cases they are talking about the same phenomenon in slightly different terms. Th
ey also recognize non-epistemic goals in scientists, and both argue that the existence of these goals doesn't necessary mean that science is irrational.
With all these similarities, their understandings of progress and rationality contrast remarkably
(note1).
For Laudan, there is no logical relationship between a research tradition and each theory in the tradition, therefore research tradition doesn't have empirical content. On the other hand, Kitcher's
practices have definite structure. Because it contains explanatory schemata, a practice will have concrete empirical content.This difference makes another more important difference. While in Laudan's view the unit of progress is not research traditions b
ut individual theories, Kitcher's scientific practices themselves are supposed to progress. In Laudan's view research traditions also evolve (95-100), but there is no criteria to judge the replacement of fundamental premises directly. We have indirect c
riteria by progress of theories in the tradition, but a tradition does not entail any theory. This means that we cannot know whether a change in a research tradition is good or bad before we see the theories it produces. There are few restrictions in how
we can modify the basic premises. As a result of this difference, Kitcher thinks that progress is cumulative (his "broadly progressive" process is obviously cumulative), and Laudan says that progress is non cumulative.
Kitcher himself men
tions these points. According to Kitcher, Laudan's argument has several defects. When we think that the units of progress are theories, many components in Kitcher's practice (paradigm of experimentation or such a thing, see p.74) drop out (89); Laudan's a
rgument cannot appreciate the accumulative explanatory structure because he sees the progress "at a level of details" (112n); Laudan cannot explain the nature of puzzle solving(130n), and so on. I think some of them are correct critiques, but
these are minor points.
One of the strongest objections to Laudan comes from the argument that his argument can be easily relativistic (130n, 157-160). As we saw here, there is an underdetermination about how we modify a research tradition (Kitcher s
hows this point more clearly using Laudan's later book; 158). If we want to avoid this underdetermination, we need a methodology outside the tradition. Laudan tried to defend rationality, but without enduring goals throughout every traditions in science,
it is puzzling how he can establish rationality. "The obvious answer is that there are enduring goals that are not represented in Laudan's official picture and that the rules of rationality formulate good strategies for attaining these goals" (
159). And Kitcher says if Laudan admits this answer his argument on progress is almost the same as Kitcher's (159). I think this analysis is correct, but the opposite is also true. That is, if Kitcher fails to show there are "enduring goals" and
"the rules of rationality" formulating "good strategies", Kitcher's argument is no better than Laudan's. (There is another argument on similar point in the note on p.130. Here Kitcher argues that Laudan's notion of puzzle solving shou
ld be (a) dependent on the notion of truth, (b) relativistic, or (c) dependent on the notion of warrant. Laudan will deny (a) and (b), but as for (c), there is a problem how to understand the notion of warrant without the notion of truth. This argument o
f Kitcher's, however, has a similar problem as the argument above. For Kitcher's argument works only when we can acquire the notion of truth without the notion of warrant. If he cannot show this, he get the same problem as Laudan have.)
As for "e
nduring goals", actually Laudan has a vague idea about it (at least in the book I talked about). For him, the aim of science is to solve problems (PIP 124). So his point should be that this enduring goal has no regulative power over what rationality
we should use. The set of problems and methodology to solve problems always change. Of course the best way to solve problems depends on what problems one tries to solve. Therefore when we discard some problems as insignificant and add some new problems
, the best methodology should also be changed. Thus scientific traditons don't necessarily share the concrete criteria of rationality. In short, Laudan doesn't have concrete methodology outside traditions because he is sure there is no such thing. This is
relativism. But does Kitcher really have an advantage here? For Kitcher, the aim of science is to find significant truth (AS 157). This is also a vague idea. If he cannot show that this vague idea has something to do with concrete standards of rationalit
y, or at least show that it can work as a regulative rule for choosing rationality, this enduring goal is not very helpful.
This problem can be answered if he can answer the next question; do we have effective rules of rationality which formulate g
ood strategies for attaining significant truth? If we have, we have enough reason to use them, and in this decision the aim of science actually works in choosing a concrete rule. His answer may be we should follow (ES). As far as we obey (ES), our practic
e will progress (almost by definition of ES). Progress means that we are proceeding to significant truth (this is also almost by definition). But here is something fatally wrong with this argument. In this picture, if we don't know where is significant tr
uth, we cannot use (ES) at all. As we saw above, his argument to connect actual progress and reality is not so successful. We can see many historical examples of progress but we cannot know if these progresses are also "progresses" in Kitcher's
sense.
There is a way to reduce this predicament. Kitcher gives us an alternative understanding of realism (171-173). In this Kantian understanding, realism is our desideratum to unify the accounts of the world. natural kinds can be interpreted as pr
edicates which are stable during the proceeding of practices. In this understanding, Kitcher should show only some process proceeds toward a stable status. In other words, he should show there are cumulative processes in science. If he is successful in th
is task, he marks clear advantage to Laudan, because, as I showed above, Laudan's progress is non-cumulative. Thus, hereafter our attentions are concentrated on whether Kitcher is successful in showing there are several methods to attain cumulative progre
ss.
His introduction of "broadly progressive" process make this task easier, because he don't have to show strict progressiveness. But this is not so easy a task. First, even if we can follow (ES) or some version of it, it does not assure we c
an progress. His standards say nothing about how high is the success ratio of the best process. Roughly speaking, if this is more than a half, we will progress; if not, we will regress. There is also a doubt about its efficiency as a criterion. He revises
(ES) several ways for real use, but some of them are problematic. For example he introduces a revision to count not only frequency of success but also significance of success (190n). This revision increase the arbitrariness in choosing according to (ES).
Anyway his argument on (ES) is too abstract to assess.
His argument for compromise model is also incomplete. He seems only to combine anti-rationalistic conditions (C1)-(C4) and rationalistic decision making (C5). He should justify this combination.
How can science be progressive in Kitcher's view?
His argument on concrete methods how to achieve progress appears in the last part of AS, chapter 7 and chapter 8. According to him, chapter 7 deals with individuals, and
chapter 8 deals with communities (9), and this suggests that these chapters are parallel. But it seems to me that chapter 8 has nothing to do with progress as it is, and it needs help from chapter 7 to have anything to do with progressiveness. To apprecia
te this point, first we take a look on the argument in chapter 8. In this chapter he uses very technical proofs, but the points he makes are simple. His concern here is how rational individual scientists can cooperate to optimize individual and social gai
ns.
First he deals with the problem of authority. He shows epistemically sullied agents (who want to become the first discoverer of something) will rationally rely on other scientists' results under certain conditions. He puts forward several techni
cal arguments on the relationship between innovation, authority, community and individual scientists. In the latter part of this chapter he deals with division of cognitive labor and theory choice. To attain new discovery, cognitive diversity is desirabl
e for community (under some situations), and sullied agents tend to attain more diversity than epistemically pure agents. Local autocracy (352) can help diversity as well.
This conclusion has direct application to theory choice. When we have two conf
licting theories, and both of them are plausible to a certain extent (this condition is C4 in his compromise model), the community will prefer the cognitive diversity to avoid the risk of overlooking the correct theory. This time he elaborates his pictur
e by adding other variables, such as cognitive variation (this is C3) and non-democratic consensus making (this is C1). Tradition effects (368) and elitism (383) allegedly can have good effects for theory choice.
In the course of this argument, he sh
ows how many seemingly negative characteristics of scientific community can play a positive role in progress of science. This is an important contribution. He seems to think he has been talking about progressiveness of community. For example, he says the
minimal contribution of these arguments is "to rebut the notion that one can infer directly from the existence of social pressures and nonepistemic motivations the conclusion that science does not advance in the fashion described earlier in this bo
ok" (388). In a sense this conclusion is right, but we should be careful what he actually showed here. His arguments depend on the utilities for the individuals and the community. When he talks about theory choice, he is just maximizing utility for
community. He talks about the probabilities scientists get correct result, but here he don't talk about how they reach the conclusion. It may maximize utility, but this does not mean it is progressive. Progressiveness depends on what method do they use. T
hus this argument is incomplete without filling in this blank. These necessary arguments can be found chapter 7.
In chapter 7, first he argues for the reliability of observation. Despite theory ladenness of observation, opponents can agree with what
is observed. His argument for observation seems to me a good one. Next he continues to defend a couple of methods in science, namely eliminative induction and escape tree.
Eliminative Induction goes as follows (237-242); First, we have a significant
question, such as, "How do the A's exemplify D(B)?"; A stands for a natural kind, for example, species of bird. D(B) is determinable B, for example, a property being blue. To answer this question, first we make a complete list of background c
onditions which can affect D(B). For simplicity, assume there are only two background conditions, namely X, Y. Here we can make one main theory and four counter theories (strictly speaking, there are 16 exclusive candidates, but I prefer simplicity).
main theory = All A's are B's
counter theory 1 = When X and Y, A's are not B's
counter theory 2 = When X and not Y, A's are not B's
counter theory 3 = When not X and Y, A's are not B's
counter theory 4 = When not X and not Y, A's are not B
's
In principle, we can eliminate these counter theories by checking each combination of conditions. In this case, we only have to observe four examples. Of course it is always possible that we overlook some relevant background conditions. This is ca
lled residual doubt (244). To eliminate this doubt, usually we collect more than four instances. Note that the question and the list of background conditions are both provided by prior practice.
Escape tree is used when we have underdeterminacy proble
ms. We have inconsistent statements, such as A; if A then B; not B (note 2). In this situation, we nevertheless want to maintain a unified account. This is called the Duhemian predicament (249). Prior practice g
ives us some constraints, so we cannot arbitrarily change one of the conflicting statements. He proposes the "escape tree" model to get the determinate solution for this situation. The escape tree goes as follows;
(1) Take away any one state
ment
(The remaining statements should be consistent, though there are some losses)
(2) Respond to the losses in the way
(i) consistent with remaining statements
(ii) compensate for the losses
(iii) conform to the constraints
(3) If
we cannot find any response that meets (i)-(iii),
pick up another statement and repeat (1)-(2)
(4) If we cannot find any response that meets (i)-(iii) to any statement,
(i) modify the constraints, or
(ii) leave open question
He does n't explain how to choose between the last two options, but his comment on significance and truth (261) suggests the way. If we feel the present explanatory schemata are significant enough, we can leave the inconsistency as it is, and repeat the process ( 1)-(3)(note3).
How can these two methods be used in scientific practice? Scientists are supposed to choose their work according to epistemic and nonepistemic goals. After scientists take sides, they may pur
sue each theory following eliminative induction and escape tree, and when one side gives the solution to their problems, the debate ends. This combination of chapter 7 and chapter 8 agrees with his compromise model. His argument in chapter 8 prepares (C1)
-(C4) in that model, and if we can justify these methods in chapter 7 by (ES), this meets the requirement of (C5).
Here are some questions how to evaluate this argument.
(1) With non-epistemic values, can scientists really follow eliminative induc
tion and escape tree? (In other words, are C1-C4 and C5 are consistent?)
(2) Do eliminative induction and escape tree have high success ratio (meet ES)?
Question (1) is important because this is the point where anti-rationalists will disagree. If
scientists have non-epistemic goals, the possibility is they will modify these pure epistemic rules to get their favorite answers. For example, sullied agents will ignore the constraint of practice to become the first discoverer. They will generalize the
ir conclusion without example, they will change explanatory schemata arbitrarily (without predicament), and so on. Why can we expect they will attain a purely epistemic consensus described in C5? Fortunately, Kitcher seems to be able to answer this challe
nge. He can introduce the distinction between context of discovery (and pursuit) and context of justification. The final goal of sullied agents is to make the discovery accepted by the community. Therefore as far as the community has purely epistemic cri
teria for assessment, we can expect that individual scientists will obey it. However, there is another question. Why can we expect the community has purely epistemic criteria? In chapter 8, Kitcher postulates the community prefer correct theory without ar
gument. Maybe to reply to this question we should consider the relation between epistemic progress and human flourishing. This question is beyond the purpose of this paper, but anyway the burden of proof is on Kitcher.
Question (2) intends to assess
not only these two rules but also ES itself. For if his own examples fail to meet ES, this standard itself should be questioned (of course I am thinking about relaxed versions of ES -- he admits original ES is too demanding).
First, How about eliminat
ive induction? Does this method have high success ratio, or, tend to result in some progress? Eliminative induction is used inside an explanatory schema, so it is unlikely to make conceptual or explanatory progress. It produce answers to significant quest
ions, so maybe it is useful to make significant statements. The generalizations by eliminative induction have some cumulative character -- as far as the explanatory schema lasts. When the schemata have changed, the appropriate answer will also change. S
o the theories should be transformed into appropriate styles. There should be something retained by this transformation (if not, it is not transformation), but the generalized theories themselves are not retained. Thus I am very skeptical that this metho
d will meet ES.
Next, How about the escape tree? This method seems to have strong tendency to attain explanatory progress. For if a solution has explanatory losses, escape tree will not admit the solution; therefore the resulting practice has at least
as much explanatory power as prior practice. There may be better method, but it does not matter when we assess with relaxed versions of ES (these versions do not require optimize the success ratio). However this strong character of escape tree model can
be the weak point at the same time. It is possible that in many occasions escape tree ends up with open question (and this is not progress in any sense). In this case, the success ratio of escape tree will be very low, and therefore will not meet ES. Thi
s is an empirical question, so I do not have a clear conclusion in this point (I am sorry if you expected a clear answer). To summarize, the answer to question (2) depends on the usefulness of escape tree model. This is something I can not answer here. Bu
t anyway we have enough reason to doubt this method have high success ratio.
After all, I think we can conclude Kitcher's rationality is vacuous as it is. We cannot find a good example for the process which meets (ES). His compromise model needs furt her supporting argument for its possibility. Without these supplementation, we should postpone the evaluation on Kitcher's promise that (ES) leads us to significant truth (or, at least, cumulative progress). This means that Kitcher doesn't answer his own critique to Laudan.
In this paper, I compared Kitcher's view and Laudan's view on progress and rationality. My conclusion is Kitcher failed to show that his theory has advantages to Laudan's without supplementation. < br> My argument is devoted to criticize Kitcher, but as is obvious from my reason to do so, I don't intend to defend Laudan. I just show Kitcher is no better than Laudan. There are many problems Laudan should also answer. For example, Laudan give us no re ason why problem solving is desirable as a aim of scientific community. He may answer that he is just describing the aim of scientific community, but in that case we can not derive normative conclusion such as rationality from this aim ( it is a kind of n aturalistic fallacy). He didn't answer how we can avoid underdetermination Kitcher points out (though Kitcher himself faces the same problem).
(1)Of course there is also a relate
d contrast between Kitcher's realism and Laudan's anti-realism, but this ontological debate is not my point here; after all, even if Laudan admits a kind of realism, most of his points on progress and rationality can be retained.
(2)As Duhem-Quine these shows, the situation is more complicated, but for this illustration the complication is not necessary.
(3)By the way, is his argument a good reply to underdeter
mination? He shows how scientists can obey a determinate rule in this situation, so in this sense he answers; but this is not what people worry about. The problem is that there are too many models similar to the escape tree. Some will start from observati
on statements, some from general laws, some from constraints themselves, and will go through similar processes as the escape tree model. There will be many answers according to where we start, and in what way we proceed. With these many variations, we fac
e another underdetermination. Thus Kitcher's answer doesn't resolve the problem.
Kitcher, P. (1993) The Advancement of Science: science without legend, objectivity without illusions, Oxford University Pre
ss.
Laudan, L. (1977) Progress and Its Problems: toward a theory of scientific grouth, University of California Press.