recovering inquiry: an epistemological approach to argument
TRANSCRIPT
Recovering Inquiry: An Epistemological Approach to Argument
Mark Weinstein, Montclair State University
My goal in this paper is to present an alternative to the traditional concern among
argument theorists with particular arguments and dialogical exchanges. The alternative offers an
image of how argumentation about significant issues in science and social policy may be
recovered and their reasonableness ascertained. The core theoretic contribution is an account of
warrant strength in terms of a logically differentiated network of support based on a
metamathematical model of emerging truth (MET), which permits the epistemological basis of
argumentation within inquiry, seen over time, to be understood and assessed (Weinstein, 2013). I
see this as broader and deeper than the usual concern with argument reconstruction, which
focuses on particular arguments presented or engaged in, and is often concerned with unstated
assumptions and other gaps in the argument structure seen in light of formal or informal
standards of argument adequacy, for example, deductive validity, inductive soundness, premise
adequacy and relevance.
I take recovering inquiry to require something that is overlooked in the logical and
dialogical perspectives. For me recovering an argument is bringing forth the reasons that in
retrospect can be seen to be compelling. Recovery, then, is seen by me to require the uncovering
of the supporting epistemological structure of the argument, how the shifting availability of
evidence and theory over time yields the epistemological power of a prevailing point of view.
The key epistemological properties seen in retrospect by recovering the structure of successful
inquiry enable us to look prospectively at ongoing debates, attempting to identify the properties
that support the likelihood of epistemological adequacy in the future.
Recovery in my sense requires that we look at the ongoing argumentation within inquiry
for epistemic power, rather than for rhetorical effectiveness, or logical soundness. We look for
structural properties that point to the power of the warrants of an argument as they come to be
better understood over time The perspective advanced here looks to the long-term epistemic
outcome of an inquiry process, which often includes multiple arguments, reflects alternative
perspectives and includes changes in relevant empirical evidence and theoretic understanding.
The MET enables us to recover the epistemological structures that warrant the acceptance or
rejection of arguments put forward as inquiry progresses.
The MET was developed in response to the apparent epistemological adequacy of
chemistry as a prototype of effective inquiry. It is based on three intuitive epistemological
principles of theory adequacy abstracted from physical science: consilience, that is, the increase
of empirical evidence over time; breadth, the range of phenomena to which a theory increasingly
applies; and depth, the ability of the theory to be reinterpreted in terms of broad ranging and
abstract explanatory theories (theoretic reduction). Successful reductions, for example, the
atomic theory as applied to known chemical processes, increase the level of detail of empirical
generalizations (consilience), as in the enormous increase in detailed understanding of chemical
processes characteristic of modern chemistry. They reach out to new areas to which the theory
can be applied (breadth), for example, the extension of chemical principles to organic substances.
And they offer a unifying explanatory account that shows an increasing field of generalizations
to be interpretable in terms of more abstract principles (depth), for example, the application of
quantum theory to chemical phenomena.
The MET addresses argument in the large, showing how complex disputes in inquiry can
be understood in terms of the strength of warrants seen within a network of support that changes
over time. The construction admits approximate models of theories in light of the prevailing
standards in the field, but requires that models be increasingly adequate, and that sequences of
such models form chains of increasing length. Such progressive model chains (consilience) and
the increase in the number of such chains (breadth) are the hallmark of an epistemologically
effective line of inquiry. Similar constructions reflect theoretic reinterpretations, that when
resulting in progressive sequences of reducing theories, mark an approach as ontologically
significant. Argumentation is adaptive and essentially convergent. Successful inquiry is seen as
large-scale convergence over time. Truth is defined in terms of the maximally successful inquiry.
These key desiderata are reflecting in the formalism in the MET, presented in abbreviated
form in the Technical Appendix, below. The three basic epistemological terms are formally
characterized in MET part I: Consilience: the increase of empirical evidence over time (MET,
1.2). Breadth: the range of phenomena to which a theory increasingly applies (MET, 1.3). Depth:
the ability of the theory to be reinterpreted in terms of broad ranging and abstract explanatory
theories, theoretic reduction) (MET, 2 through 2.4). A dialogical model of warrant strength based
on the MET is part II in the appendix. The relation between MET and truth through ontological
commitment is part III.
The MET was initially applied to the development of the Periodic Table and the
persistence of Prout’s hypothesis in the face of counter-evidence (Weinstein, 2011). The
application of the MET to physical science is relatively non-controversial; more controversial is
its application to other sorts of inquiry. In what follows I will briefly sketch out two applications:
multi-theoretic arguments in response to scientific racism and speculative claims in cognitive
science.
Scientific Racism
The efforts of Franz Boas and his followers to reject the racism endemic after WWI are
well documented (Barkan, 1992). What is salient for the discussion here is the commitment by
Boas and his followers to historical particularism (Harris, 2001, pp. 250ff). This raises deep
problems for the perspective that that MET affords, for the MET is based on a clearly nomothetic
science, physical chemistry, and the notion of warrant strength that is at the core of its
application to argumentation requires just what an ideographic approach to anthropology
eschews.
As Harris and others have indicated, although the focus in anthropology was on rich
descriptions of cultural practices, such descriptions are not immune to a theoretic overlay.
Although Boas focused on culture as the major determinant of human understanding and
behavior those that followed offered theoretic overlays that included Freudianism (Benedict and
Mead), functionalism (Malinowski), structuralism (Levi-Strauss) and materialism (Harris). Each
of these adds support to anti-racism by positing theory-based cross-cultural commonalities that
unify rather than differentiate groups. It is such theoretic accounts that support the relevance of
the rich descriptions that Boas’ approach requires, contributing over-arching theoretic constructs
in terms of which the description can be understood. The wealth of descriptive data sets the
standards for the adequacy of theoretic accounts, supporting or challenging the theory as
increasing detail is brought forward. This is captured by the initial insight that the MET affords,
that theory must is supported by a sequence of models that satisfy the theory or are increasingly
adequate to what the theory implies (MET, 1.2).
It is the perspective of the MET that rich descriptions, however essential in the
understanding of human culture, are preliminary and that the adequacy of anthropological
understanding requires that generalizations by put forward and tested by their ability to warrant
increasingly adequate accounts of the phenomena. Such generalizations determine competing
theoretic perspectives. The MET attempts to offer an indication as to how such competing
perspectives may be evaluated. The view put forward is that generalizations take their strength
from the breadth and depth with which they connect networks of explanatory structures (MET,
1.3). This is evident in the large explanatory frameworks, for example, structuralism and
materialism, which are put forward as an overlay on the accepted generalizations in a field
(Harris, 2001, pp. 634ff). It is not surprising that such large frameworks are the subjects of
debate within anthropology, but if we are not to succumb to relativism, the logical appraisal of
such theoretic frameworks is essential. This will be clearer as we move from anthropology to
biology, where the possibility of reduction to deep explanatory theories becomes possible.
The original biological arguments for distinguishing races reflected both the taxonomical
model that would prove a mainstay of biology and a crude mixture of physical and cultural
attributes that signaled the complexity of the attempt to make sense of human diversity. As the
classification morphed into a hierarchical “chain of being,” analogies were sought with the
physical appearance and behavior of animals, especially supposedly man-like apes (Jordan,
1968). But the fundamental distinction based on skin color remained essential, expanded in later
race theorists to include physical measures of the ratio of body parts that supposedly signal both
aesthetic and functional superiority (Gould, 1981, pp. 127ff). The main focus of continuing
research was the measurement of skulls as an indicator of both innate intelligence and
evolutionary status (Gould, pp. 73ff). The underlying theory was that of polygeny, a proto-
biblical doctrine of the separate creation of the various races championed by Louis Agassiz and
reflected in the work of Samuel Morton (ibid., pp. 42ff) . Morton’s work and the work of Paul
Broca have been criticized by Gould, who identified both the empirical and theoretical failing of
the work (Gould, op. cit.) but even if Gould’s work is challenged as to the details the findings
suffer from clear lack of theoretic connection to intelligence, that is with effective cognitive
functioning (Graves, p. 47). The MET points to the underlying problem. Even if we accept such
measurements, whatever their accuracy, it did not result in a continuing and increasingly
effective research programs in craniometry. That is, it did not result in a progressive model chain
(consilience, MET, 1.2). Moreover it did not result in additional chains of models elaborating the
relationship of craniometry to aspects of psychological functioning (breadth, MET, 1.3). The
work of Broca attempting to extend craniometry to, for example, hierarchies of occupational
status and social class resulted in a dead-end (Gould, pp. 82ff).
The issue of polygeny adds additional interest. If polygeny is correct, whether in its
biblical version or in an evolutionary tree which differentiates the races in terms of distinct
evolutionary history, the biological theory of distinct races is at least possible, and possible
supportive of racist interpretations. But that requires that their be some notion of race that is
sufficiently similar to the socially constructed notion, now determined primarily by self-
identification and the evolutionary model. We will briefly explore this in terms of recent work in
the genetic basis of paleoanthropology, below, but for now suffice it to say that the MET shows
why it is a scientific possibility. The reinterpretation of sets of empirical data under an
overarching theory, reduction (MET, 2 through 2.4) offers a warrant potential enormous power,
for its confirmatory yield is potentially a function of the joint confirmation of all of the more
empirical theories that it reduces. The analogy with chemistry is telling. Seeing the chemistry of
gases, the structure of crystals, the electro-chemical properties of substances, the dynamics of
fluids, the tensile strength of solids and the bio-chemistry of living things as all interpretable in
terms of the underlying structure of molecules and atoms is the basis for the priority of physical
chemistry as a paradigm of truth. The improbability of such a grand unification of disparate
sciences, each with their own history, methodology and evidence attests to its enormous
epistemological power on both intuitive and Bayesian grounds. And so we cannot rule out the
potential of a evolutionary version of polygeny despite the failure of the initial research program
in its name.
Before we return to genetics, a development of the late 20th century, we see what the
MET has to say about the central role of psychometrics as the scientific warrant for maintaining
and engineering racial and social difference, the basis for eugenics and social policies that gave
scientific support for segregated schools, sterilization and ultimately genocide. The basic
apparatus for the psychometric support of eugenics was the availability of intelligence tests
beginning with the work of Alfred Binet (Gould, pp. 146ff). The results of intelligence testing
was related to Mendelian genetics by H. H. Goddard, which supported the program in eugenics
to inhibit the reproduction of mentally defectives (ibid., pp. 163ff) and given additional social
relevance by Lewis Terman and especially by R. M. Yerkes in his testing of army recruits during
World War I. Gould sees Terman to have recanted seeing that “mean differences are too small to
provide any predictive information for individuals” (ibid., p. 192) and sees Yerkes to have made
fundamental methodological and statistical errors (ibid., pp. 199ff) and to have disregarded the
plausible interpretations of his findings as evidence of the radically different social and
educational environment that that characterizes the individuals tested (ibid., pp. 217-222). All of
which can be seen as a failure of the psychometricians to develop progressive model chains. But
there is a deeper criticism that Gould employs in challenging the Mendelian interpretation of the
results of testing. Gould asks, “Could the plethora of causes and phenomena grouped under the
rubric of mental deficiency possibly be ordered usefully on a single scale?” (ibid., p. 159). His
answer is that it could not, seeing “mental retardation, specific learning disabilities caused by
neurological damage, environmental disadvantages, cultural differences” as all plausible
contributors to the score on an intelligence test and so making the hypothesis of a single genetic
cause unreasonable (ibid., pp. 159-160). In terms of the MET, the genetic interpretation of the
testing data is not a useful explanatory model for the phenomena. Cyril Burt and his followers
would beg to differ.
Burt, Charles Spearman, Arthur Jensen and the authors of The Bell Curve (Hernstein &
Murray, 1994) now armed with an array of tests that indicate aspects of cognitive competence
rely on the statistical technique of factor analysis to argue for a univocal notion of intelligence, g,
which reduces the correlations among the various test to one major component, and thus
accounts for the apparent variety of abilities in terms of one unifying, possible genetic,
underlying cause. The argument is complex and relies on technical discussions of factor analysis
and the openness of the technique to alternative constructions (ibid., pp. 214-215) but the deep
issue is the reification of factors. Does the statistical fact, whatever it is, have ontological
consequences? This is the problem of reification, are statistical artifacts real properties of the
existent things that they measure? (ibid., pp. 238-239, pp. 250-252; pp. 268-269). Does the
statistical fact, whatever it is, have ontological consequences?
The MET offers a perspective on that issue. The MET gives an account of ontological
commitment internal to a scientific theory (Weinstein, 2002). The technical details are in the
Technical Appendix, part III, but the upshot of the formalism is that we commit ourselves to an
ontology in terms of the domain of the intended model of our most successful theories, where
that commitment is captured by the Quinean aphorism, to be is to be the value of a variable.
Thus, factor analysis can tell us what our theories must account for, that is if g is supported by
the statistical evidence, then an explanatory theory of what g is must be forthcoming. If, for
example, we think human cognitive capacity is a function of the central nervous system the
neurophysiological correlate for g must be discovered. If, in addition, we think the
neurophysiological correlate is genetic, as opposed to the development of the brain under the
impress of environmental stimulation, the underlying genetic structure must be identified. Thus
factor analytic results, not matter how reconstructed, only have ontological significance when
they are cashed out in a theory of increasing empirical adequacy (MET, 1.2) and breadth of
application (MET, 1.3, 2.3) in the best case a theory whose breadth enables it to incorporate and
reinterpret the range of empirical models and explanatory theories in its own terms (MET, 2.4).
The perspective on scientific racism the MET affords is, however, equivocal, for a theory
of racial differences can be seen as supported by a genetic account of human diversity based on
recent work in paleoanthropology (Wade, 2006). Modern genetics is among the broadest and
deepest scientific theories available outside of the physical chemistry, making the contribution to
traditional evolutionary biology that led to the grand synthesis that is modern evolutionary
theory. Genetics satisfies the desiderata of consilience, breadth and depth, and its depth is
profound, furnishing the bridge between biology and physical chemistry. But whatever the
theoretic warrant of the work of paleoanthropologists, concerns with the application of a genetic
based racial theory face the charge of racism, which arise due to socially sensitive social
applications of modern genetics to medicine and criminal justice (Krimsky & Sloan, 2011).
Nevertheless, tracing the genetic variation of human populations subsequent to the migration
from Africa distinguishes groups of people in a fashion that reflects, to some extent, the
traditional division of humans into races. For example the ‘chromosomal tree’ based on the Y
chromosome, passed from father to son, includes the mutation M173 that distinguish populations
in Europe from those that populate the Americas, characterized by M242 as distinguished from
the inhabitants of West Eurasia with M170 and have the parent population in Africa
characterized by M168. This is reflected in the distribution of mutations based on mitochondrial
DNA passed along the female line that include distinct lineages that are geographically
differentiated, albeit in complex ways (Wade, 2006, pp. 56ff).
Genetic theories are powerful potential reducers. They track evidence as diverse as
migrations identified archeologically through settlement ruins, the persistence of tool making
cultures, the dispersion of language families and patterns of resistance to disease. Genetic
theories sit in deep and broad reduction chains. Genetic claims are supported by chemical
analysis and are consistent with known principles of biology and physiology and so fit within a
detailed and comprehensive inquiry project that accommodates a wide range of explanatory
structures independent of their application to the biology of race. But even if the science of
genetics is strongly supported by the theories that surround it, the question still remains as to
whether it is a strong reducing theory in respect of its application to explaining human
differences relevant to notions of race.
The question is whether race as genetically defined explains race as socially constructed,
based on self-attribution in light of traditional racial categories. The first thing to notice is that
race is socially constructed in various ways, particularly in countries in the Western Hemisphere
with a long history of racial mixing. People descended from two different racial classifications
can be grouped in two ways, hypodescent, where the designated race has lower status, as when a
bi-racial individual of African descent is designated as African-American or hyperdescent, which
designates according to the higher status racial designation, common in Brazil (Krinsky and
Sloan, 2011, pp. 246ff). This, in itself, makes it unreasonable to expect genetics based racial
theory to fit the traditional categories of race. The social construction of race is not a coherent
categorical system; it reflects different social histories and encompasses competing distinctions,
including physical characteristics, language use, religion and ethnicity.
Racial groups can, however, be differentiated by genetic mutations, tracing both male and
female lines, which have geographic distributions that reflect socially constructed racial
distinctions to some extent. An interesting case concerns recent attempts to identify genetic
markers for racially targeted and effective medical interventions. Reliance on racial distinctions
in such contexts becomes more supportable and more capable of resisting social and moral
complaints about their possible misuse. Policy is a negotiation between fact and value, but we do
this at our peril unless we get the facts right. The MET addresses the scientific basis by focusing
on the large-scale structures that warrant the empirical evidence. This is also evident in the
perspective it affords on understanding developing inquiry projects such as cognitive science.
Cognitive Science
Cognitive science, like the chemistry of Dalton, begins with deep theoretic concepts that
serve as potential reducing theories for newly acquired, but relatively impoverished, empirical
data. This is captured in the MET by admitting increasingly adequate approximation relations
among the models rather than requiring exactitude (consilience rather than validity). From the
perspective of the MET it is not surprising that theories are inadequate to their models in many
ways, and that the debates among proponents of competing points of view may remain
unresolved as inquiry progresses. Early chemistry included theoretical constructs that were
inadequate to the phenomena. But the descriptions of the phenomena reflected both empirical
and conceptual flaws (Weinstein, 2011). So, for example, data sets for the relative proportions of
chemical components were subject to the vagaries of inadequate measurement (Scerri, 2007, p.
40). And even as measurements improved, empirical models of chemical reactions could not
possibly be given an adequate theoretical account until the discovery of isotopes (Scerri, 2007, p.
58). When applied to physical chemistry, the MET looks to the developing of the network of
ideas over time and the interplay of empirical evidence and theoretic modeling. This exposes an
essential aspect of inquiry that illuminates the potential of speculations within cognitive science.
Cognitive science exhibits enormous potential breadth and depth. It looks at the wide
range of human cognitive activities and draws its substance from two powerful sources, the
natural sciences that support neurophysiology and the logical basis of computer science. Natural
science increases the evidentiary weight of cognitive science proposals to the extent that they can
be connected to well-established theories, ranging from anatomy to physics. The roots of
cognitive science in logic and computer science offers a level of mathematical construction that
differentiates as needed by particular fields in inquiry. The MET is neutral in terms of the details.
All the MET requires is that there be a comparative assessment of models over time, that
sequences of models are progressive, increasingly adequate to the phenomena over time
(consilience, MET 1.2).
Breadth of concern is perhaps the most apparent characteristic of cognitive science. The
Cambridge Handbook of Cognitive Science lists 8 related research areas that reflect different
aspects of human cognition, standardly construed, and extends the reach of cognitive science to
include animal cognition, socially mediated cognition, evolutionary psychology and, most
essential, the bridge between cognitive science and the rest of physical science, cognitive
neuroscience (Frankish and Ramsey, 2012). Each of these is a going concern, and none of them
is free of difficulties. Yet in all cases there is a sense of advance, of wider and more thoughtful
articulation of theoretical perspectives that address a growing range of concerns. The MET (1.3)
offers a logical account of why breadth is a telling epistemological attribute, crucial for
evaluating the structure of support that warrants confidence in the value of the enterprise and its
ultimate vindication as ontologically significant (MET, part III).
The study of memory serves as an indication of the progressive nature of cognitive
science. The cognitive architecture of memory, the discussion between short and long-term
memory has been understood for some time. With the additional concept of working memory the
model for understanding memory encoding and retrieval was in place. Elaboration and
controversies still abound, but the basic physiological structures though which memory can be
physically impaired have been identified. Additional details and functional analyses have been
postulated, for example the distinction between declarative and episodic memory and the
relationship between recollection and familiarity has been explored both experimentally and
physiologically. More adequate knowledge of brain anatomy connects different levels of
analysis. Such studies include accounts of cognitive deficits as a function of physical
deformation of the brain, as well as explanations of successful cognition in terms of underlying
brain mechanisms. For example, fMRI studies offer detailed accounts of visual memories in
terms of neurophysiological descriptions of the visual cortical areas (Ranganath et. al., 2012). In
terms of the MET, increasingly adequate empirical models of memory (consilience, MET 1.2)
form related sequences of models (breadth, MET 1.3), which reinterpret cognitive processes in
terms of higher-order neurological models (depth, MET, 2).
The MET offers a different perspective on research on reasoning and decision making.
There are a variety of models for understanding reasoning and decision making derived from
empirical and behavioral studies, expanding upon the logic derived paradigms familiar from the
work of Wason and Johnson-Laird (1972), Nisbett and Ross (1980) and others (see Oaksford et.
al., 2012). In contrast to this empirical work, speculative efforts look beyond logic to broader
considerations, offering possibilities of deeper understanding than normative-based empirical
paradigms can afford. Damasio (1995) has resuscitated the connection between reasoning and
emotion and Thagard & Aubie, (2008) see a connection between cognition, emotions and levels
of awareness. If such complex understanding can be connected with underlying structural and
functional models of the brain there is the possibility of significant progress beyond more simple
logic-based models. At the present, such efforts rely on computer simulations using virtual
neurons in the absence of detailed knowledge of the brain. But advances in the understanding of
the brain’s structure and function offer the possibility of deep reductions that have ontological
significance. This explains the pervasive interest in cognitive science as a materialist theory of
mind despite philosophical objections to the contrary. The MET enables us to for the details.
Thagard and Aubie draw upon both neurophysiology and computer modeling. This
enables both theoretic breadth (MET, 1.3) and the possibility of increasing adequacy (MET, 1.2),
even if the latter is no more that computer simulations of simplified cognitive tasks. They cite
ANDREA, a model which “involves the interaction of at least seven major brain areas that
contribute to evaluation of potential actions: the amygdala, orbitofrontal cortex, anterior
cingulate cortex, dorsolateral pre-frontal cortex, the ventral striatum, midbrain dopaminergic
neurons, and serotonergic neurons centered in the dorsal raphe nucleus of the brainstem”
(Thagard and Aubie, 2008, p. 815). With ANDREA as the empirical basis, they construct
EMOCON, which models emotional appraisals, based on a model of explanatory coherence, in
terms of 5 key dimensions that determine responses: valance, intensity, change, integration and
differentiation (pp. 816ff). EMOCON employs parallel constraint satisfaction based on a
program, NECO, which provide elements needed to construct systems of artificial neural
populations that can perform complex functions (p. 824ff. see pp. 831 ff. for the mathematical
details). This points to the potential power of their approach. Computer models, even if gross
simplifications, permit of ramping up (MET 2.1). A logical basis with a clear mathematical
articulation has enormous potential descriptive power as evidenced by the history of physical
science (MET 1.3).
Damasio (2010) has a similarly ambitious program. He begins with the brain’s ability to
monitor primordial states of the body, for example, the presence of chemical molecules
(interoceptive), physiological awareness, such as the position of the limbs (proprioceptive), and
the external world based on perceptual input (extroceptive). He construes this as the ability to
construct maps and connects these functions with areas of the brain based on current research
(pp. 74ff.). This becomes the basis for his association of maps with images defined in neural
terms, which will ground his theory of the conscious brain.
Given that much he gives an account of emotions elaborating on his earlier work, but
now connecting emotions with perceived feelings. As with the association of maps and images,
Damasio associates emotions with feeling and offers the following account: “Feeling of emotions
are composite perceptions of (1) a particular state of the body, during actual or simulated
emotion, and (2) a state of altered cognitive resources and the deployment of certain mental
scripts” (p. 124). As before he draws upon available knowledge of the physiology of emotional
states but the purpose of the discussion is not an account of emotions per se, but rather to ground
the discussion of memory, which becomes the core of his attempt at a cognitive architecture (pp.
339ff.). The main task is to construct a system of information transfer within the brain and from
the body the brain. The model is, again, mediated by available physiological fact and theory
about brain function and structure and so permits a broad range of models (MET, 1.3). The main
theoretic construct in his discussion of memory is the postulation of ‘convergence-divergence
zones’ (CDZs), which store ‘mental scripts’ (pp. 151ff.). Mental scripts are the basis of the core
notion of stored ‘dispositions,’ which he construes as ‘know-how’ that enables the
‘reconstruction of explicit representation when they are needed” (p. 150). Like maps (images)
and emotions (feelings) memory requires the ability of parts of the brain to store procedures that
reactivate prior internal states when triggered by other parts of the brain or states of the body.
Dispositions, unlike images and feelings are unconscious, ‘abstract records of potentialities’ (p.
154) that enable retrieval of prior images, feelings and words through a process of reconstruction
based in CDZs, what he calls ‘time-locked retroactivation’ (p. 155). CDZs form feedforward
loops with, e.g. sensory information and feedback to the place of origination in accordance with
coordinated input from other CDZs via convergence-divergence regions (CDRegions) by
analogy with airport hubs (pp. 154ff.). Damasio indicates empirical evidence in primate brains
for such regions and zones (p. 155) and offers examples of how the architecture works in
understanding visual imagery and recall (pp. 158ff.), again offering a potentially broad range of
models (MET, 1.3) each of which can, in principle, offer the possibility of increasing articulation
over time (MET, 1.2).
The result of all of this is an attempt, as the title of the book suggests, to construct a
brain-based theory of self, which building on what he has developed so far distinguishes three
stages, the proto-self “a neural description of relatively stable aspects of the organism....
spontaneous feeling of the living body,” the core self, “which connects the body to the external
world through “ a narrative sequence of images, some of which are feelings” and an
autobiographical self “when objects in one’s biography generate pulses of the core self that are,
subsequently, momentarily linked in a large-scale coherent pattern” (p. 192).
Damasio like Thagard and Aubie offer speculative models that reference current
physiological knowledge, rely on concepts from computer science and information theory and
bypass the deep philosophical issues that are seen by many to create an unbridgeable gap
between the mental and the physical short of deep metaphysical reorientation (Chalmers, 1996).
Yet, whatever the ultimate verdict on these two authors, the rich program in cognitive science
persists and has a strong appeal. The reason is the potential strength of the warrants, that is to
say, if such models prove to be correct the epistemic force of the warrants that support them will
be enormous, swamping the force of alternative approaches that rely on, for example,
psychological evidence alone. This requires a more careful look at the perspective that the MET
provides.
The MET determines a hierarchy of epistemic adequacy in terms of models and chains of
models viewed over time. (MET, 1-1.3) Each level of adequacy supports correlative levels of
warrant strength (MET, part II). The level of warrant strength has consequences both for the
acceptance of the theory and for its power to resist counterexamples (see Weinstein, 2013,
chapter 4 for the dialectical details and a related adaptive logic.). For a theory to have sufficient
warrant to be taken seriously it must reflect its intended models in that it either holds in the
models MET, 1.1, a) or is increasingly adequate to the evidence it strives to explain (MET, 1.1,
b). But the models in which it holds, whether exactly or with better approximations over time are
frequently a small set of the available concerns potentially within the scope of the theory.
Looking at the history of the periodic table we find a similar pattern. Theoretic models held for
small subsets of the known chemical elements and theoretic approximations to empirical data
were typical. But as the research program persisted more and more chemicals were brought
under the scope of explanatory models and approximations of empirical data improved as both
theoretical and the experimental understanding was refined.
Given the claims of both Thagard and Aubie and Damasio to base their models on
accepted facts about brain function, if proved correct, the accounts, however speculative meet the
first test and so are warranted at a minimal level. That is their views capture aspects of the brain
or they approximate accepted knowledge to a degree that is close enough to merit consideration.
If they are close enough approximations, we look to their progress as they refine their models
and as knowledge of brain function increases. If the approximations are, in general, progressive
as defined in the MET the speculations are seen as increasingly adequate. Adequacy in light of
neurological facts is compelling and increasing adequacy is a sign of the fecundity of the
theoretic approach as chains of linked models progress (MET, 1.2).
Cognitive scientists who connect cognition with other brain functions, that like cognition,
require and mediate information across systems (for example, physiological control and
emotions) add empirically relevant models of essential brain functions, so the theory is not
merely more adequate to its models, but there is an increasing range of models to which it
applies (MET, 1.3). Again this is typical of the history of the periodic table and was a predictor
of its potential strength as the research program flourished.
Both Thagard and Aubie and Damasio take synoptic approaches and offer models which
cross the boundaries of brain functions, offering generalizable schemes for neural architecture.
This shows enormous potential for breadth. The far-ranging interests of cognitive science lend
prima facie force to any reasonable attempt at articulating a neurophysiological account of core
cognitive functions that might account for increasing aspects of the field. The wide range of
empirical and theoretic studies characteristic of the field points to enormous potential breadth for
anybody who gets it right, mirroring the history of the periodic table. Physical chemistry was
initially concerned with gases; over time, independent areas of studies, ultimately including the
entire range of physical substances, were incorporated under the basic concept of periodicity, as
the basic ideas were reorganized around theoretic advance and more adequate empirical
evidence. The result is a massive unification of the entire field of physical chemistry, arguable
the most successful inquiry project in human history. Whatever the challenges, the epistemic
payoff of a correct cognitive science is enormous, whence the power of the field despite its many
problems
Tying cognitive science to neurophysiology gives an evolving empirical basis with
warrants tied to the underlying structures of physiology. Physiological understanding is
increasingly grounded in foundational sciences such as biochemistry and electro-chemistry. The
empirical basis is necessary but it is the foundational knowledge that ultimately has the more
powerful evidentiary force. Reducing neuroanatomy to a functional neurophysiology is the
pathway to physicalism. Claims within physical science have the most powerful warrants,
supported by networks of evidence at the highest level of articulation and affording enormous
explanatory depth (MET, part II). Speculative talk about c-fibers reflecting what little was
known about the physical correlates for mental episodes (in this case pain) was deemed worthy
of decades of philosophical discussion just because the possibility of reducing the mental to the
array of physical knowledge grounded the mental firmly within the scientific worldview
(Hardcastle, 1997). The MET attempts to capture this in the model of reduction (MET, 2-2.4).
But, unlike much of the discussion of the mind-body problem, which was concerned with
identity, the MET sees reduction through identification. The reduction relation in the MET does
not seek identities, but rather tracks the growing identification of one theory by the replacement
of models, that is it offers reinterpretations of aspects of theories when appropriate model
relations hold (MET, 2). But what is most important is the history of reductions, which using the
model of physical science is often partial and approximate. As we can reinterpret more and more
phenomena in terms of a more basic theory our confidence in the warrants that result increases,
first as a function of the adequacy of the reduction, then the increasing depth of the reduction
(MET, 2.1), the increase in theoretic adequacy in light of the reduction (MET, 2.2), the increases
in theoretical reach as the various reductions mutually reinforcement refinement in theory in
light of symmetries between the various theories in light of the over-arching reducing theory
(MET, 2.3) and finally the increase in scope across large areas of inquiry as the reducing theory
captures networks of theories (MET, 2.4). It is on the basis of such a history of progress that
ontological claims are warranted (MET, part III) and is the basis for the view that scientific
materialism is the most plausible candidate for what the world is really made of.
5. CONCLUSION
If cognitive scientists are successful in modeling cognitive behavior in terms of brain
processes, and if, as is becoming more evident, a wide range of psychological processes are
implicated in cognition, possible co-extensive with the range of phenomena identified with so
called folk psychology, the possibility of a scientific basis for the mind becomes more than
philosophical speculation. Whether cognitive scientists will succeed remains to be seen. Whether
a grounding of the mental in the physical will satisfy philosophers is even more uncertain,
especially as phenomenology becomes a favored perspective among philosophers. But short of a
wholesale disregard of science, perhaps in the name of some heir of post-modernism, cognitive
science has a potential for epistemic adequacy that transcends the arguments that support
particular claims. It is this last point that indicates the importance of the perspective the MET
affords to the theory of argument.
The MET enables the epistemic structure that supports the potential strength of warrants
to be seen and in doing so indicates a perspective that looks at the large epistemological
structures that give the ultimate strength to ongoing inquiry. Elsewhere I have speculated that the
approach of the MET has relevance beyond science to political, legal and ethical argument
(Weinstein, 2009b, 2013. chapter 5). It seems apparent to me that a number of essential social
and political concerns could benefit from seeing arguments as networks of reinforcing points of
view, networks that give differential support as a function of the breadth and depth of the
warrants that they contain. My attempt here to offer and analysis of the complex arguments
brought forward against the scientific arguments for racism common until WWII is one such
attempt. Arguments concerning global warming seem another obvious arena for my approach.
Sorting out the epistemic force of the scientific, economic and environmental warrants that
characterize competing points of view seems essential if we are to understand the issues and
develop coherent policies.
Whether argument theorists will take my proposals seriously remains to be seen. The
formalism of my approach need not be a barrier. The metamathematics gives formal substance to
the basic concepts, but it can be seen as formal metaphor for the three essential and hopefully
intuitive epistemological desiderata: consilience breadth and depth. But with or without the
formalism, my approach requires a repositioning of the theory of argument. Arguments, on my
view, are not limited to games with winners and losers or even to attempts at persuasion. Rather
they are seen as the ongoing rational process of clarifying essential issues. Such attempts
whether in scientific inquiry or in social policy transcend particular argumentative exchanges and
project backward onto the available evidence and project forward as additional evidence
accumulates and positions are revised. The MET shows how such a network can be envisioned
and how complex substantive arguments can the understood.
TECHNICAL APPENDIX
The Model of Emerging Truth (MET)
Part I:
1. A scientific structure, TT = T, FF, RR (physical chemistry is the paradigmatic example) where
T is a set of sentences that constitute the linguistic statement of TT closed under some
appropriate consequence relation and where FF is a set of functions F, such that for each F in FF,
there is a map f in F, such that f(T) = m, for some model or near model of T. And where RR is a
field of sets of representing functions, R, such that for all R in RR and every r in R, there is some
theory T* and r represents T in T*, in respect of some subset of T.
A scientific structure is first of all, a set of nomic generalizations, the theoretic commitments of
the members of the field in respect of a given body of inquiry. We then include distinguishable
sets of possible models (or appropriately approximate models) and a set of reducing theories (or
near reducers). What we will be interested in is a realization of TT, that is to say a triple T, F,
R where F and R represent choices from FF and RR, respectively. What we look at is the history
of realizations, that is, an ordered n-tuple: T,F1,R1,...,T,Fn,Rn ordered in time. The claim is
that the adequacy of TT as a scientific structure is a complex function of the set of realizations.
1.1. Let T be a subtheory of T in the sense that T is the restriction of the relational symbols
of T to some sub-set of these. Let f be subset of some f in F, in some realization of TT. Let
T1,...,Tn be an ordered n-tuple such that for each i,j (i<j,) Ti reflects a subset of T modeled
under some f at some time earlier than Tj. We say the T is model progressive under f iff:
a) Tk is identical to T for all indices k, or
b) the ordered n-tuple T1,...,Tn is well ordered in time by the subset relation. That is to say, for
each Ti, Tj in T1,…Tn (ij2), if Ti is earlier in time than Tj, Ti is a proper subset of Tj.
1.2. We define a model chain C, for theory, T, as an ordered n-tuple m1,...,mn, such that for each
mi in the chain mi = di, fi, for some domain di, and assignment function fi, and where for each di
and dj in any mi, di = dj; and where for each i and j (ij), mi is an earlier realization (in time) of T
then mj.
Let M be an intended model of T, making sure that f(T) = M for some f in F ( for some
realization T, F, R and T is model progressive under f. We then say that C is a progressive
model chain iff:
a) for every mi in C, mi is isomorphic to M, or
b) there is an ordering of models in C such that for most pairs mi, mj (j i) in C, mj is a nearer
isomorph to M than mi.
This last condition is an idealization, as are all similar conditions that follow. We cannot assume
that all theoretic advances are progressive. Frequently, theories move backwards without being,
thereby, rejected. We are looking for a preponderance of evidence or where possible, a statistic.
Nor can we define this a priori. What counts as an advance is a judgment in respect of a
particular enterprise over time best made pragmatically by members of the field. We are engaged
in rational reconstruction where logically clarity trumps descriptive adequacy, in presentation,
but where descriptive adequacy is still at the heart of the intuition.
1.3. Let C1,...,Cn be a well ordering of the progressive model chains of TT, such that for all i,j
(i > j), Ci is a later model chain than Cj. TT is model chain progressive iff C1,...,Cn is well
ordered in time by the subset relation. That is to say each later model includes and extends the
models antecedent to it in time.
2. We now turn out attention to the members of some R in RR. The members of RR represent T in
T* in respect of some subset of T, k(T). Let k1(T),...,kn(T) be an n-tuple of representations of T
over time, that is, if
i > j, then ki(T) is a representation of T in T* at a time later that kj(T). We say that TT is
reduction progressive iff,
a) k(T) is identical to Con(T) for all indices, or
b) the n-tuple is well ordered by the subset relation.
2.1. We call an n-tuple of theories RC = T1,...,Tn a reduction chain, and T1,...,Tn a deeper
reduction chain than j-tuple T1,...,Tj, iff n>j and for all i,j there is a ri in Ri such that ri
represents Ti in Ti+1 and similarly for Ti and further for all Tk (k< j) Tk is identical in both
chains Note, the index i must be different from the index j, since if i=j, there is no Ti+1.
2.2. We call a theory reduction chain progressive iff T iff for an n-tuple of reduction chains RC1,...,
RCn and for each RCi (i<n), RCi+1 is a deeper reduction chain than Rci.
2.3. T is a branching reducer iff there is a pair (at least) T and T* such that there is some r and r*
in R’ and R*, respectively, such that r represents T in T and r* represents T* in T and neither T
is represented in T* nor conversely.
2.3.1. B = TT1,TT2,...,TTn = T1, F1, R1,T2, F2,R2,...,Tn, Fn, Rn is a reduction branch of TT1 iff
T1 is a branching reducer in respect of Ti, and Tj (i 2; j 3 for i,j n)
2.4. We say that a branching reducer , T is a progressively branching reducer iff the n-tuple of
reduction branches B1,...,Bn is well ordered in time by the subset relation, that is, for each pair
i,j (i>j) Bi is a later branch than Bj, that is, the number of branching reducers has been increasing
in breadth as inquiry persists.
Part II:
The core construction is where a theory T is confronted with a counterexample, a specific model
of a data set inconsistent with T. The interesting case is where T has prima facie credibility, that
is, where T is at least model progressive, that is, is increasingly confirmed over time (Part I, 1).
A. The basic notion is that a model, cm, is a confirming model of theory T in TT, a model of data, of
some experimental set-up or a set of systematic observations interpreted in light of the prevailing
theory that warrants the data being used. And where
1) cm. is either a model of T or
2) cm is an approximation to a model of T and is the nth member of a sequence of models ordered
in time and T is model progressive (1.1).
B. A model interpretable in T, but not a confirming model of T is an anomalous model.
The definitions of warrant strength from the previous section reflect a natural hierarchy of
theoretic embeddedness: model progressive, (1.1), model chain progressive (1.3) reduction
progressive (2), reduction chain progressive (2.2), branching reducers (2.3) and progressively
branching reducers (2.4). A/O opposition varies with the strength of the theory. This defines
levels of warrant strength, W1 through W6, respectively. Anomalies face a dialectical burden as
a function of the strength of warrants. We require that a warrant be model progressive at least,
that is there is evidence that is either stable or getting better.
So, if T is merely model progressive, W1, an anomalous model is type-1 anomalous, if in
addition, model chain progressive, W2, type-2 anomalous etc. up to type-6 anomalous for
theories that are progressively branching reducers, W6.
P1. The strength of the anomaly is inversely proportional to dialectical resistance, that is,
counter-evidence afforded by an anomaly will be considered as a refutation of T as a function of
strength of T in relation to TT. In terms of dialectical obligation, a claimant is dialectically
responsible to account for type 1 anomalies or reject T and less so as the type of the anomalies
increases.
P2: Strength of an anomaly is directly proportional to dialectical advantage, that is, the
anomalous evidence will be considered as refuting as a function of the power of the explanatory
structure within which it sits.
P*: The dialectical use of refutation is rational to the extent that it is an additive function of P1
and P2
Part III.
3. Let TTT be an n-tuple TT1,...,TTn of scientific structures seriously proposed at a time. Let
T1,F1,R1,...,Tn,Fn,Rn be their respective realizations at a time. We say that a set of models M,
M = {m1, m2,...,mn} is a persistent model set iff
a) M = m1=d1,f1, m2= d2,f2,..., mn = dn,fn and for all i, j di = dj, or
b) M is a model in a set of ordered subsets of TTT, such that the sequence is well ordered in time
by the subset relation.
3.1 M is an ontic set for TTT, that is, M is a set of models that are putative ontologies for M in
that their common domain persists under reduction relations over time.
3.2. We say that an ontic set O is a favored ontic set iff,
a) O is the set of intended models of a theory T is the first member of a progressive
reduction chain. (O is thus the ontic set of all of the theories in the chain.)
b) the members of the reduction chain are themselves reduction progressive.
c) T is a progressively branching reducer.
3.2.1. Notice that the set consisting of an ontic set and the sets that it generates (the set of sets
under the reduction relation), form a persistent model set.
3. TT is progressive if:
a) TT is model chain progressive
b) TT is model progressive
c) TT is reduction progressive.
3.1. We call T a progressive reducer if:
a) T is reduction chain progressive
b) T is a progressively branching reducer.
3.2. We say T is a favored reducer, if:
a) TT is progressive
b) T is a progressive reducer.
3.3. T is a most favored reducer if T is a maximally progressive reducer, that is, T is the nth
member of a reduction chain such that for all Ti, (i<n) Ti is a favored reducer. (Notice, T is not
reduction progressive, since it stands at the head of the longest reduction chain).
3.3.1. The set O, of ontic models of T, is thus, a favored ontic set in respect of every Ti (i<n) in
the reduction chain.
3.3.2. If T is a most favored reducer, and O is its favored ontic set than O*= {m1,...,mn} of
models mi in O is the ontology of scientific structure TT.
3.3.3. A truth predicate for TT can then be constructed in fairly standard Tarskian as ‘s is true’
for s in T and T in TT, iff O*||-s where O* is the ontology of TT.
References
Barkan, E. (1992). The Retreat of Scientific Racism. Cambridge: University of
Cambridge Press.
Chalmers, D. (1996). The Conscious Mind. New York: Oxford University Press.
Damasio, A. (1995). Descartes’ Error: Emotion, Reason and the Human Brain. New York:
Harper.
Frankish K. and Ramsey, W. (2012). The Cambridge Handbook of Cognitive Science.
Cambridge: Cambridge UP.
Gould, S. (1981). The Mismeasure of Man. New York: Norton.
Graves, J. (2008). The Emperor’s New Clothes. New Brunswick, NJ: Rutgers University Press.
Hardcastle, V.G., 1997, “When a Pain Is Not,” Journal of Philosophy, 94: 381–409.
Harris, M. (2001). The Rise of Anthropological Theory. New York: Rowman and Littlefield.
Hernstein, R. and Murray, C. (1994). The Bell Curve. New York: Free Press.
Jordan, W. (1968). White Over Black. Chapel Hill: University of North Carolina.
Krimsky, S. and Sloan, K. (2011). Race and the Genetic Revolution. New York:
Columbia University Press.
Nisbett, R and Ross, B. (1980). Human Inference. Englewood Cliffs, NJ: Prentice Hall.
Oaksford, M, Chater, N. and Stewart, N. Reasoning and decision making. In Frankish and
Ramsey (2012).
Ranganath, Libby and Wong (2012). Human learning and memory. In Frankish and
Ramsey (2012).
Scerri, E. R. (2007). The Periodic Table: Its Story and Its Significance. New York:
Oxford University Press.
Thagard, P. and Aubie, B. (2008). Emotional consciousness: A neural model of how
cognitive appraisal and somatic perception interact to produce qualitative
experience. Consciousness and Cognition, 17, 811-834
Wade, N. (2006). Before the Dawn. New York: Penguin.
Wason, P. and Johnson-Laird, P. (1972). The Psychology of Reasoning: Cambridge, MA:
Harvard UP.