This preview shows page 1. Sign up to view the full content.
Unformatted text preview: Psychology, Philosophy, and Cognitive Science:
Reﬂections on the History and Philosophy of
Abstract: This article critically examines the views that psychology ﬁrst came into
existence as a discipline ca. 1879, that philosophy and psychology were estranged in
the ensuing decades, that psychology ﬁnally became scientiﬁc through the inﬂuence
of logical empiricism, and that it should now disappear in favor of cognitive science
and neuroscience. It argues that psychology had a natural philosophical phase (from
antiquity) that waxed in the seventeenth and eighteenth centuries, that this psychology
transformed into experimental psychology ca. 1900, that philosophers and psychologists
collaboratively discussed the subject matter and methods of psychology in the ﬁrst
two decades of the twentieth century, that the neobehaviorists were not substantively
inﬂuenced by the Vienna Circle, that the study of perception and cognition in psychology did not disappear in the behaviorist period and so did not reemerge as a
result of artiﬁcial intelligence, linguistics, and the computer analogy, that although some
psychologists adopted the language-of-thought approach of traditional cognitive
science, many did not, and that psychology will not go away because it contributes
independently of cognitive science and neuroscience. 1. A Science of Psychology?
Psychology has been self-consciously trying to be a science for two hundred
years, give or take ﬁfty. In the meantime it has developed a variety of laboratory techniques, collected much experimental data, shown some theoretical
development, and undergone changes of opinion about whether its primary
object of study is mind or behavior. Has it made its way to sciencehood? Some
have thought psychology became scientiﬁc by freeing itself from philosophy
near the end of the nineteenth century, while others make it wait for behaviorism and positivism. A few recent thinkers believe that psychology can remain
scientiﬁc only by becoming something else: neuroscience, cognitive science,
or those and more.
*This is the ﬁfth in a series of Millennial pieces that have been specially commissioned by Mind &
Language. Each contributor has been allowed choice as to topic and approach, being asked only
to present a personal view of issues that he or she sees as being important in the present context
of cognitive studies.
The author dedicates this essay to the memory of his brother, James L. Stanton, 1946–2001.
Address for correspondence: Department of Philosophy, University of Pennsylvania, Logan
Hall, Room 433, Philadelphia, PA 19104–6304, USA.
Email: hatﬁeld phil.upenn.edu
Mind & Language, Vol. 17 No. 3 June 2002, pp. 207–232. Blackwell Publishers Ltd. 2002, 108 Cowley Road, Oxford, OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA. 208 G. Hatﬁeld These various positions on psychology’s sciencehood offer speciﬁc claims
about the subject matter, methods, and explanatory adequacy of an autonomous discipline of psychology. The psychologists themselves have frequently
afﬁrmed that they became scientists when Wundt and others broke free from
philosophy by establishing psychological laboratories and conducting genuine
experiments on mental phenomena, using techniques and apparatus imported
from physics and physiology. In a phrase once popular, psychology became a
science by rising from the ‘armchair’ of speculation and uncontrolled introspection, and entering the laboratory to undertake controlled observation and
measurement.1 During the second quarter of the twentieth century some psychologists and philosophers argued that these initial efforts had not resulted in
a truly scientiﬁc psychology, because of a persistent mentalistic infection. Talk
about mind or mental experience was, on this view, inherently subjective and
unscientiﬁc. Accordingly, psychology was able to become a science only with
the rise of behaviorism, as promulgated ﬁrst by J.B. Watson under the inﬂuence
of J. Loeb and other physiologists, and rendered genuinely scientiﬁc when C.
Hull and B.F. Skinner adopted the scientiﬁc methodology of logical positivism
or logical empiricism.
The recent authors who believe that psychology should become something
else grant that the discipline was propaedeutic to the scientiﬁc study of cognition, but deny that such a discipline should or will remain autonomous.
Instead, neuroscience, cognitive science, linguistics, and others will partition
its domain. Michael Gazzaniga, reﬂecting on the layout of the psychology
building at Dartmouth College in Hanover, New Hampshire, concludes that
there is no overarching discipline of psychology to hold together the various
ﬂoors, devoted to social psychology, cognitive science, and cognitive neuroscience. In his view, ‘psychology itself is dead’ (1998, p. xi). The cognitive and
perceptual part of psychology has gone over to the ‘evolutionary biologists,
cognitive scientists, neuroscientists, psychophysicists, linguists, [and] computer
you name it’ (p. xi). Howard Gardner sees the cognitive areas of psychology
merging with artiﬁcial intelligence, to form ‘the central region of a new uniﬁed
cognitive science’ (1985, p. 136). The remaining discipline of psychology will 1 It is typical to characterize early experimental psychology as simply introspective. But in fact
early experimentalists such as Wundt (1902, sec. 3) and Titchener (1912) were critics of
uncontrolled introspection. They relied on subjects’ (or self-) reports of perceptual experience, but they collected such reports in controlled circumstances in which the stimulus was
known and the subject was focused on matching one stimulus to another, discriminating
between stimuli, or estimating a value. The notion of skilled introspective analysis of sensations as developed by Titchener did come in for heavy criticism, but it was not the only
or the primary notion of introspection in American psychology (on introspection in Wundt
and Titchener see Danziger, 1980). And once the Gestaltists immigrated (Kohler, 1929;
Koffka, 1935), phenomenally-based reports of experience, distinct from the analytical introspection of Titchener, were well represented on the American scene. Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 209 be limited to various applied activities (in the clinic, the schools, and industry),
and a few subject areas such as personality and motivation.
Who is right about whether, when, and how psychology became a science?
That is a question for the history and philosophy of psychology, a blended
mixture of historical scholarship and philosophical analysis. As a way of
reﬂecting on mind and language in the past century or so of psychology, I
want to consider some key themes and episodes in the history and philosophy
of the experimental psychology of perception and cognition. These are the
purported founding of scientiﬁc psychology ca. 1879, the relation of psychology to philosophy then and in the subsequent ﬁfty years, and the rise of
cognitive science in the 1960s and 70s. These themes and episodes highlight
aspects of the relations between scientiﬁc psychology and philosophy, as well
as, in recent times, between psychology and cognitive science. In the end, I’ll
have my own prediction on whether psychology is going away.
2. The Founding of Psychology as a Scientiﬁc Discipline
Around the world psychologists celebrated the centenary of their discipline in
1979, a date chosen to mark 100 years since Wundt set up a laboratory at the
University of Leipzig (having previously used his instruments at home). The
celebratory literature was replete with knowing concessions that the precise
date was somewhat arbitrary—one could look earlier to Fechner’s experiments,
or discuss whether Wundt’s was really the ﬁrst laboratory—but nonetheless
accepted the date as approximately correct.
In my view the claim that psychology was created anew as a scientiﬁc
discipline in 1879 or thereabouts is profoundly misleading. It obscures the
disciplinary and theoretical continuity of the new experimental psychology
with a previous, natural philosophical psychology. And it goes together with
a story of rapid antagonism between philosophy and psychology at century’s
turn, which itself seriously misrepresents the state of play between philosophers
and psychologists at the time.
Other natural sciences,2 such as physics, biology, and chemistry, do not
have a founding date. They mark milestones in their histories, but the study 2 With this phrase I render the question of when psychology became a science as the question
of when it became a natural science. Most of the practitioners of the ‘new psychology’ of
the late nineteenth century considered their discipline to be, or to be trying to be, a natural
science, though not all did. Wundt had at ﬁrst thought of psychology as a natural science
(in the 1860s), but later treated it as belonging to the ‘sciences of spirit’ (Geisteswissenschaften),
though with some methodological commonalities with natural science. In the ﬁrst half of
the twentieth century the University of Pennsylvania (1938, p. 44) listed as one area of
required instruction the ‘Natural Sciences’: botany, chemistry, mathematics, physics, psychology, and zoology. The classiﬁcation of psychology under natural science continues at
Penn today. This classiﬁcation was shared with Aristotle, Kant, and William James (see Hatﬁeld, 1997). Blackwell Publishers Ltd. 2002 210 G. Hatﬁeld of the general properties of the natural world, of living things, and of chemical
phenomena stretch back to the medieval period and, at least for physics and
biology, back to Aristotle (though the term ‘biology’ is an eighteenth-century
coinage). The milestones they mark often involve radical transformations of
the disciplines. These transformations may have brought into existence the
modern form of the science, and so the modern discipline may see itself as
having really sprung from these more recent achievements. But historians of
the sciences, and practitioners themselves when they are thinking historically,
allow that the study of their subject matter, considered generally, existed in
Consider physics. As the study of the general properties of natural things
—including living things and rational or minded beings—physics was a standard discipline in Aristotle’s scheme of knowledge, a scheme that set the educational and theoretical framework for European universities from the thirteenth to eighteenth centuries. Physics was distinct from mathematical
astronomy, a discipline that extended back to the ancient world and was represented by Ptolemy’s Almagest. After some preliminary work by Nicolaus
Copernicus, Johannes Kepler, Rene Descartes, and others, in the late seven´
teenth century Newton propounded a new scheme of mechanics which could
serve as the basis for astronomy, through the inverse-square law of gravitation
and Newton’s laws of motion. Newton’s discoveries led to the development
of modern physics, which leaves living things to biology, and minded things
to psychology. So Newton’s achievement marks a kind of beginning that
physicists can recognize. It was a big success that got them going on a track
that led to where they are now (even though Newton’s science was eclipsed
by further radical transformations).
But it would be silly to say that Newton’s work marked the beginning of
physics as a discipline, for the discipline existed under that name from Aristotle’s time on. In its Aristotelian form it was not, like astronomy, a mathematical
discipline; physics dealt with the ‘natures of things’, while separate mathematical sciences such as optics and astronomy described things using geometry.
Some of the preliminaries to Newton involved bringing mathematics to bear
on a wider range of natural phenomena, as in Galileo’s ‘new’ science of motion
(which actually had medieval roots). But not all the conditions that made
Newton’s achievement possible involved mathematical applications or discoveries. There was also a conceptual background, ultimately derived (in function if not in content) from the ‘natural philosophical’ approach of Aristotle.
Natural philosophy was, even in Newton’s time, another name for physics. It
meant what it says: philosophy about nature. It was ‘philosophical’ in that
there was discussion of the basic classiﬁcation of natural things into kinds,
characterization of the properties of those kinds, and an explanatory framework
invoking those kinds and their relations. This sort of natural philosophical work
was performed by Descartes, who sought to replace the Aristotelian natural
philosophy through the bold vision of a uniﬁed physics of heaven and earth Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 211 governed by a few laws of motion (see Hatﬁeld, 1996). Historically, Newton’s
mathematical physics was a mathematized correction of Descartes’ physics.
The point of this historical excursus is to introduce the notion of a natural
philosophical background to a recognizably modern, mathematics-using,
experiment-generating scientiﬁc discipline. Psychology also has a natural philosophical background. Its ultimate source is again Aristotle, through his De
anima or ‘On the Soul’. The Latin word anima translates the Greek psyche,
which is the root for the modern term ‘psychology’. For reasons that remain
obscure, but may have to do with the awkwardness of the noun form ‘animistics’ as opposed to ‘psychology’, the discipline slowly changed its name from
de anima studies to psychology across the seventeenth and early eighteenth
centuries. But the study of the functions of the mind or soul was continuous.
In the early period, Aristotelean psychology included the study of vital as well
as sensory and cognitive functions (‘soul’ for Aristotle simply meant vivifying
principle—though in fact Aristotle and his followers spent most of their time
on the sensory and cognitive functions in the works entitled ‘On the Soul’).
Cartesian psychology, by contrast, included only the sensory, cognitive, and
affective dimensions of mind: those that are available to human consciousness.
This narrowing of the subject matter to the contents of consciousness took
hold, and became a standard way of delimiting psychology in the eighteenth century.
Psychology, like physics, had a long history as a natural philosophical discipline. This history exhibits a second parallel with the history of physics. Just
as Newton uniﬁed the mathematical treatment of nature with physics as the
study of nature in general to form the modern discipline of physics, Descartes
and his followers (with some anticipation by Ibn al-Haytham) brought a previously mathematical discipline into psychology: the discipline of optics, considered as the theory of vision (see Hatﬁeld, forthcoming). The ﬁeld of optics,
which had been cultivated by Ptolemy (2nd century) and advanced by Ibn alHaytham (11th century) included geometrical theories of size and distance
perception, to account for size constancy and distance perception. Descartes
developed this theory in his work on Dioptrics, but also in the psychological
portions of his Treatise on Man. His followers incorporated these perceptual
theories into the parts of their textbooks on natural philosophy that treated
the mind or soul (see Hatﬁeld, 1995).
For reasons that have not been fully explored, calls for a more empirical,
physics-emulating psychology came thick and fast around 1750. The Swiss
naturalist Charles Bonnet published his Essai de psychologie in 1755. GuillaumeLambert Godart published his Physique de l’ame, or ‘Physics (i.e., natural
philosophy) of the Soul’ in 1755, and Johann Gottlob Kruger published his
Experimental-Seelenlehre, or ‘Experimental Psychology’, in 1756. Each of them
called for application of the empirical attitudes found in other branches of
science, whether physiology, botany, entomology, or Newtonian science, to
the domain of the mental. None of them, least of all Godart, proposed to treat Blackwell Publishers Ltd. 2002 212 G. Hatﬁeld the psychological from a materialistic standpoint. Their proposals were not
about the ontology of the mind, and none of them saw any contradiction in
allowing that the mind might be immaterial even while proposing to study its
states and processes through observation and experiment. Kruger, especially,
advocated what later was called ‘empirical dualism’ (in Schmid 1796, pp. 189–
90), the position that the phenomena of mind and body form distinct subject
matters, whatever the underlying ontology might be. In any event, by the
second half of the eighteenth century there was a thriving market for textbooks
in psychology, so that even at that time one can ﬁnd the standard prefatory
remark offering justiﬁcation for ‘yet another’ textbook of psychology (Bonnet,
1755; Kruger, 1756). Further, these textbooks referred to some early quantitat¨
ive experiments on sensory phenomena, including Patrick d’Arcy’s (1765)
determination of the rate of the decay of retinal stimulation, through a method
of positive afterimages.
Psychology’s long natural philosophical phase saw a slow growth in mathematically framed theories (as of the constancies) and in quantitative empirical
work. Throughout the nineteenth century there were more and more calls to
make psychology into a true natural science. These were attempts to change
it from a branch of the old natural philosophy (or from a ‘moral’ discipline, as
in Scotland and early nineteenth-century France) into a self-standing empirical
science using up-to-date laboratory methods derived from physics and physiology. These results culminated in an explosion of new psychological laboratories during the 1890s and early 1900s (especially in North America, see Hilgard, 1987, ch. 1). This stunning rapidity fed the idea that a totally new science
had been ‘founded’.
In fact the new science that took hold and developed in the late nineteenth
and early twentieth centuries was a transformation of the old science (or old
‘natural philosophy’). This can be seen in the theoretical continuity between
the old and new, especially in the core area of sense perception. The theories
of spatial perception (including size and distance perception) of Wundt and
Helmholtz take their theoretical bearings (directly or indirectly) from the earlier work of Johann Steinbuch, Caspar Tourtual, and Hermann Lotze, which
relied on the theories of George Berkeley and William Porterﬁeld, which in
turn were framed by the work of Ibn al-Haytham and Descartes (see Hatﬁeld,
1990b; forthcoming). Although the theoretical framework was reﬁned and
more fully articulated by the later generations, the basic description of size
constancy (that for near distances we perceive things as having close to their
true size, whether they are 5, 10, or 15 feet away), and the notion that constructive judgmental or associative processes underlie size constancy, were there
from the start. The phenomena of distance perception had been studied experimentally in the eighteenth century. E. H. Weber, Fechner, Helmholtz, and
Wundt successfully subjected yet further aspects of sensory perception to
measurement, and developed precise methods for studying depth perception
(see Turner, 1994). But the theoretical notions they employed were directly Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 213 continuous with previous work. Often, as in the case of James’ Principles of
Psychology, which set the framework for American psychology in the 1890s
and beyond, the continuity was openly acknowledged. James showed appreciation for the descriptive and theoretical contributions of earlier writers, including not only mid nineteenth-century writers such as Alexander Bain and the
Mills, but also the eighteenth-century German psychologist Christian Wolff
(James, 1890, 1: pp. 356–9, 409, 484–7, 651).
If all or most of this is true, then why the conventional story of psychology’s
novel founding ca. 1879? Several things explain this story, including ulterior
motives and the conﬂation of coincidents. First the coincidents. The late nineteenth century saw a reorganization of American universities as they began
graduate programs. Over the course of the century, the sciences, which had
been included in the ‘philosophy faculty’ in accordance with the Aristotelian
classiﬁcation of physics as natural philosophy, separated out into science faculties (though as they started offering the doctorate it was still called the Ph.D.,
or doctorate in philosophy). Psychology separated from philosophy proper a
little later, due to the late arrival of the notion of ‘scientiﬁc psychology’
(current in Europe from the eighteenth century) to the North American continent. Hence, the normal process of disciplinary consolidation could be mistaken in psychology’s case for a totally new founding (more on this below).
But the primary reasons are ulterior. The discipline of psychology in the
United States, Germany, and Britain retained a close connection with philosophy into the early part of the twentieth century. Although this connection
was often amicable, some experimentalists saw it as a holdover of ‘metaphysical’
and ‘armchair’ psychology from the pre-scientiﬁc days. Moreover, the experimentalists then (as now) keenly feared that applied and clinical psychology
would turn the discipline into a merely applied science. They therefore took
action to consolidate the image of psychology as at core an experimental
science, which meant distancing themselves from both philosophy and application. In the US, a ﬁrst act of distancing occurred during the 1890s, as many
departments of psychology were founded independently of philosophy. This
allowed for comparatively amiable relations between philosophy and psychology in the ﬁrst quarter of the twentieth century. But when E. G. Boring
composed his History of Experimental Psychology (1929), he was intent on deﬁning the new science through its experimental method (O’Donnell, 1979);
hence, connections with philosophy were relegated to a pre-historical phase
from which psychology proper had emerged by freeing itself from philosophy,
making it seem as though psychology were newly existent ca. 1879, instead
of merely being transformed.
3. Behaviorism and Philosophy
For Wundt, James and others psychology was the science of mental life, or of
conscious mental states. Behavior might provide a form of evidence, but there Blackwell Publishers Ltd. 2002 214 G. Hatﬁeld was no thought that psychology was aimed at predicting behavior. Wundt and
James were in general agreement that conscious mental life might be explained
via laws of mental life itself, and they both considered physiological conditions
to be relevant—though Wundt would not allow physiological causal explanations in psychology, as opposed to mental causal explanations which he did
allow (Wundt, 1902, p. 28). They further differed in that James, reﬂecting the
Darwinian functionalism of American psychology, emphasized the explanatory
power of viewing the mind teleologically, as functioning to adjust the organism
to the environment (James 1890, 1: pp. 8, 79).
In the decades after the turn of the century the focus on conscious states
as the primary object of study in psychology was challenged. Several authors
proposed that psychology should concern itself with explaining and predicting
human ‘conduct’, or the ‘behavior’ of organisms more generally. Some early
proponents of this attitude, such as William McDougall (1905, 1912), then at
Oxford prior to moving to Harvard and then Duke, and Walter Pillsbury
(1911) at Michigan, suggested that behavior should be explained using the
mentalistic vocabulary of traditional psychology, and that introspection was
among the methods to be used in discovering mentalistic explanatory facts.
This was a behavioral psychology, but without a denial of mentalism. It was
one of several forms of behavioral psychology, some of which denied the
validity of introspection but retained mentalistic terms in the description of
behavior itself, and the most radical of which sought to expunge every mentalistic concept, whether limited to behavioral description or not, from the
language of psychology.
The most famous of the early radical behaviorisms was that of John Watson,
erstwhile Professor of Psychology at Johns Hopkins University in Baltimore.
Watson was a materialist, who believed the ultimate explanation of behavior
would be physical and chemical (1913, 1914). Although in his view physiology
would be the ultimate behavioral science, in the meantime a behavioral psychology might serve to explain behavior by charting stimulus-response
relations, using Pavlov’s conditioning theory and Thorndike’s laws of effect.
Such a psychology could ﬁnally join the ranks of legitimate natural science,
because it would reject all inherently unscientiﬁc (in Watson’s view) mentalistic notions, whether introspective or descriptive of behavior itself. Stimulus,
response, and measurable bodily states would be discussed using only the
(presumably objective) vocabulary of physical description.
Watson saw himself as ﬁnally placing psychology on the road to scientiﬁc
respectability. According to typical historical accounts (Boring, 1950, p. 657;
Koch, 1964, p. 10; Leahey, 1980, pp. 303–6), the job was completed when
the neobehaviorists Tolman (1932), Hull (1943), and Skinner (1938) married
the focus on behavior and the rejection of introspective mentalism with the
methodological sophistication of Viennese logical empiricism. Although Watson’s form of behaviorism had remained metaphysical in its forthright materialism, Tolman and company allegedly rendered psychology into pure science by Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 215 adopting the (anti-metaphysical) methodological physicalism of Carnap (1932
) and Hempel (1935 ). Carnap and Hempel proposed that all
descriptions of behavior and its causes must be translatable into the language
of physics.3 Since the only afﬁrmation here concerned the proper vocabulary
of description and explanation, without endorsing ‘material mode’ claims about
ontology, logical empiricism allegedly allowed psychology to free itself of
metaphysics and to gain methodological objectivity. Accordingly, American
psychology came into its own by gaining a new rapprochement with philosophy—not with metaphysics, but with the logic of methodology, or the
philosophy of science (Smith, 1981).
There are two aspects of this standard story I want to challenge. First, it
accepts, either tacitly or explicitly, the common view that after psychology
separated itself from philosophy in the 1890s (or into the following decade),
a state of hostility existed between the disciplines until the positivistic rapprochement. Second, it misdescribes the positions of the neobehaviorists themselves, imputing connections which did not exist between their work in the
1930s and that of the logical empiricists. Indeed, the neobehaviorists all rejected
central tenets of the physicalist or logical behaviorism of Carnap (1932) and
The revisionist picture of the relation between behaviorism and logical
empiricism has been forged through the work of Ron Amundson (1983, 1986)
and, subsequently, Laurence Smith (1986). Through historical research, these
authors have shown that the neobehaviorisms of Tolman and Hull were formed during the 1920s, both having felt the inﬂuence of American NeoRealism, and Hull having developed a penchant for deductive explanation in
science from his conversations with C. I. Lewis, A. N. Whitehead, and his
study of Newton’s Principia (Smith, 1986, p 165). Hence neither was inﬂuenced in their formative years by Vienna. Subsequently, in the 1930s, each
had some contact with logical empiricism and the Unity of Science movement,
and in the 1940s and 1950s each was interpreted as having adopted the logical
analysis or the physicalism of Carnap and others. But in fact they both took
little or nothing from Vienna. Tolman did not think that behavior could or
should be interpreted in a purely physicalist language. He maintained that mentalistic notions such as goal and expectation could be used to describe and
3 Carnap put the point as follows: ‘Every psychological sentence refers to physical occurrences
in the body of the person (or persons) in question’ (1932 [1959, p. 197]). Molar behavioral
descriptions might be used in our present state of ignorance, but they are to be regarded as
coarse ways of referring ‘to systematic assignments of numbers to space-time points’; ‘Understanding “physics” in this way, we can rephrase our thesis—a particular thesis of physicalism—
as follows: psychology is a branch of physics’ (p. 197). Hempel summed up his point as follows:
‘All psychological statements which are meaningful, that is to say, which are in principle
veriﬁable, are translatable into propositions which do not involve psychological concepts, but
only the concepts of physics. The propositions of psychology are consequently physicalistic
propositions. Psychology is an integral part of physics’ (1935 [1949, p. 378]). Blackwell Publishers Ltd. 2002 216 G. Hatﬁeld explain behavior when appropriately tied to responses, or response tendencies.
Indeed, Tolman was willing to postulate intervening psychological processes
involving ‘cognitive postulations’ and ‘representations’ (possessing the marks
of intentionality) to explain the maze-running behavior of rats (1926, 1927
[1951, pp. 60, 65]). These processes were realistically conceived, but were not
framed in the language of physiology (or physics). Tolman allowed psychology
to posit explanatory intervening states described in a purely psychological language (even if it was assumed that the states had a neural realization).
Hull was also a realist, but of a different stripe. He was a materialist realist
who wanted to exclude all mentalistic notions from the description and explanation of behavior. Like Watson, he felt that the ultimate explanations in psychology should come in neurophysiological terms. But he shared with Tolman
a belief that at present it was premature to restrict the explanatory vocabulary of
psychology in that way. Although rejecting Tolman’s intentional or mentalistic
notions, he identiﬁed himself with Tolman as a ‘molar’ behaviorist, arguing
that behavior theory could progress despite the lack of knowledge in neurophysiology, and granting behavioral science its own observational and theoretical vocabulary. He was willing to introduce undeﬁned theoretical terms such
as ‘habit strength’, which he believed referred to unobservable internal states
of the organism. He hoped to eventually deﬁne such terms by using empirical
data to determine the laws of habit strength. This approach was not in step
with the ongoing Viennese accounts of theoretical terms as gaining meaning
through their role in a formal system (see Smith, 1986, ch. 7). Hull came away
disappointed by his contacts with the Unity of Science movement. As he
reached the limitations of his previously preferred Newtonian geometricalstyle of formalized theory, he looked to Woodger’s (1938) axiomatization of
biology for help. But Woodger’s adherence to the logical empiricist notions
of implicit deﬁnition for undeﬁned theoretical terms, and his unwillingness or
inability to come to terms with Hull’s realistically conceived intervening variables, led to a breakdown in their collaboration. Hull remained a materialistic
realist devoted to providing a realist interpretation of his unobservables. Nonetheless, Hull’s students and colleagues, including G. Bergmann and K. Spence
(1941), and S. Koch (1941, 1964), interpreted him as a devotee of logical
empiricism. Koch (1964) used the presumed connection between Hullian
behaviorism and logical empiricism as part of an argument to the effect that
since logical empiricism had been discredited, Hull’s behaviorism was discredited, too. This guilt by association failed to come to terms with Hull’s
own brand of materialist, realist hypothetico-deductive science (on which, see
Amundson and Smith, 1984).
Skinner differed from Tolman and Hull in rejecting their realism, a position
he cast as a rejection of theory in behavioral science. Skinner had been exposed
to Machian positivism prior to 1930, and he adopted Mach’s anti-metaphysical
inductivism, his focus on biological adjustment, and his suspicion of posited
theoretical entities. In the early 1930s Skinner had a favorable response to Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 217 Carnap’s (1932) early mention of behaviorism, and to Bridgman’s (1927) operationism. After meeting Carnap in 1936 he expressed reservations about relying
heavily on logic in the analysis of science, and subsequently put aside both
Bridgman’s operationism and logical empiricism as overly formal and physicalistic (1938, 1945). Although Skinner rejected mentalistic terms that could
not be fully translated into neutral behavioral descriptions, he did not think
behaviorist psychology should be reduced to physiology nor that its descriptions could be restated in physical language. He did allow that one goal of
science might be to discover connections among differing ‘levels of description’
(such as the neuronal and behavioral levels), but he was unenthusiastic about
the unity of science as a program. Emphatically, he considered discovery of
such connections to be no part of the science of behavior as he understood
it (1938, pp. 418, 429), thereby rejecting the physicalist vision of psychology.
He was also leery of materialism’s tendency to overlook the behavioral level of
analysis in favor of concrete physical states of the organism (1938, pp. 440–1).
The standard story of a close alliance between logical empiricism and neobehaviorism turns out to be largely a retrospective fabrication. That does not,
however, mean that there was no direct philosophical inﬂuence on the formation of behaviorism. Indeed, philosophers were involved in the formulation
and discussion of behaviorism from the very beginning. And these philosophical discussions played a formative role in the development of Hull’s and
especially Tolman’s neobehaviorism.
Nearly a century hence, when we look back at the formation of behaviorism, Watson’s original manifesto (1913), and his subsequent books (1914, 1919)
loom large. Some note might be taken of the inspiration Watson took from
Loeb’s (1900) tropistic approach to animal behavior, but the movement in
psychology is laid largely at his door.
Things seemed different, and indeed were different, at the time. When
A.A. Roback, then an instructor in psychology at Harvard, published in 1923
a book examining the recent history and present state of the behaviorist
controversy, he identiﬁed no fewer than eight separate strands of behaviorism
proper, as well as two versions of ‘psycho-behaviorism’ (mixing some of the
categories of introspective mentalism with behaviorism) and six varieties of
‘nominal behaviorism’ (fully mentalist in orientation). In reviewing the rich
literature of behaviorism, Roback looked not only to the American Journal of
Psychology and the Psychological Review, but also to the Journal of Philosophy,
Psychology, and Scientiﬁc Method and the Philosophical Review. Further, the books
and articles he reviewed were written not only by individuals who are uncontroversially considered to be psychologists, such as Watson, J.R. Kantor, M.
Meyer, A.P. Weiss, R.M. Yerkes, Tolman, Pillsbury, and McDougall, but also
by philosophers, including E.A. Singer, R.B. Perry, E.B. Holt, and G.A. De
Laguna. And the philosophers were not presented as simply reacting to the
behaviorist writings of the psychologists. J. Dewey, G. Santayana, and F.J.E.
Woodbridge were listed as ‘pre-behaviorists’, presumably for their biological, Blackwell Publishers Ltd. 2002 218 G. Hatﬁeld functionalist attitude toward the organism as reacting to the environment.
Singer (1911) was credited with having contributed ‘one of the earliest explicit
behavioristic credos (Roback 1923, p. 44).
Roback treated the philosophers as intellectual peers who contributed on
equal footing to a discussion about the possibility, characteristics, and prospects
for a behavioristic psychology. He considered psychology, once it entered the
‘experimental state’, to have ‘emancipated’ itself from philosophy (1923, pp.
97–8). But he also believed that every science, psychology included, was at
all times subject to ‘philosophical audit’, and that any science contemplating
a shift in methodology and subject matter would of necessity engage in philosophical work, self-consciously so or not.
Roback’s inclusion of the philosophers in his work was reasonable, for they
had in fact informed and shaped the intellectual debate. Simply on the numbers, philosophical interest in behaviorism must be judged signiﬁcant. Between
1911 and 1925, in the Journal of Philosophy alone there were 14 articles with
‘behavior’ or ‘behaviorism’ in the title, beginning with a discussion note on
Singer (1911). The contributors included the psychologists Watson and Tolman, and the philosophers Woodbridge and Stephen Pepper. Further, we have
seen above that two out of three of the major neo-behaviorists, Tolman and
Hull, owed a major debt to the American Neo-Realists, who included Perry
and Holt. Tolman retained strong ties to Perry, and frequently cited his work
throughout the 1920s and into the 1930s (along with the work of Pepper and
Singer). Further, Singer converted a fourth neo-behaviorist, E. B. Guthrie, to
behaviorism through his philosophy classes at the University of Pennsylvania.
Hull was converted to behaviorism by teaching two seminars on it at the
University of Wisconsin. He used Watson (1925) and Roback (1923) as texts,
and reportedly stressed ‘the philosophical background of behaviorism’ (Smith,
1986, p. 152). While he found Watson’s version of behaviorism wanting, he
believed that the shift toward a non-mentalistic, behavioral psychology could
succeed if it were better executed, both empirically and conceptually or philosophically.
This picture may seem historically anomalous, for two reasons. First, there
is a working assumption that philosophy of science, or philosophical examination of the sciences, arose in America when the logical empiricists immigrated
during the 1930s. Second, there is an even more widely held view that the
disciplines of philosophy and psychology were estranged in the early decades
of the century.
Neither of these assumptions bears scrutiny. In the period from 1890
onward there was in fact considerable discussion of the relation between philosophy and science. The main streams of American philosophy, including
Realism, Neo-Realism, and Critical Realism, all advocated that philosophy
should take the results of the sciences seriously. Of the sciences, biology and
psychology were given the most attention, though discussions of physical
science were not rare. C.S. Peirce, Dewey, James, and J. Royce each had Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 219 considerable scientiﬁc education and wrote about scientiﬁc topics, as did
Singer, A.O. Lovejoy, M.R. Cohen, and of course A.N. Whitehead. Singer
regularly taught a course in Philosophy of Science at Pennsylvania beginning
in 1896–97, and in Development of Scientiﬁc Thought from 1898–99; at
Harvard, Royce taught courses pertaining to the criticism and examination of
the special sciences (such as Cosmology or Philosophy of Nature) from 1885–
86. Columbia was something of a center for the history and philosophy of
science, study of which was encouraged by Woodbridge and Cohen, and later
(from 1930) by E. Nagel. This program had already produced, as a doctoral
dissertation in 1925, E.A. Burtt’s Metaphysical Foundations of Modern Physical
Science: A Historical and Critical Essay, which was an important stimulus to the
further study of the history and philosophy of science in France by A. Koyre
(on which, see Hatﬁeld, 1990a). The active interchange between philosophers
and psychologists over the advent of behaviorism was an expectable expression
of the prevailing naturalism and scientiﬁc interests of American philosophy in
the ﬁrst decades of the century.
The idea that philosophers and psychologists felt estranged in American
universities after the turn of the century is if anything more fully an artifact
of projective reconstruction. In the literature of the history of psychology there
has been considerable examination of the situation in German universities during the late nineteenth and early twentieth centuries, where the philosophers
and psychologists were battling over the appointment of experimental psychologists to chairs in philosophy. Such appointments were an outgrowth of the
fact that from the late sixteenth century the discipline called ‘psychology’ had
been taught by philosophers (who also taught logic, natural philosophy, moral
philosophy, and metaphysics). As psychologists began to do laboratory work,
they naturally wanted to occupy chairs with appropriate resources for setting
up a psychological institute. But the ministers of education created few new
chairs, instead ﬁlling philosophy chairs with experimentalists. Tension between
the two groups was manifested in the various charges of ‘psychologism’ bandied about during the ﬁrst two decades of the century (and catalogued in
A story of similar tensions on the American scene has been portrayed by
the historian Daniel J. Wilson (1990). He speaks of the philosophers in America
feeling ‘inferior’ in their interchanges with psychologists, making up for this
by founding their own society (the American Philosophical Association, or
APA) in 1901 so they wouldn’t have to meet with the psychologists anymore
(who had founded their APA, the American Psychological Association, in
1892), and then going into a crisis in the ﬁrst decades of the twentieth century
over their relation to science. But Wilson presents precious little evidence for
this crisis. He also seems to have little understanding of the basic tenor of
philosophy in every age, for he suggests (p. 125) that philosophy’s failure at
discipline-formation was manifest in the lack of broad substantive agreement
on matters philosophical among all practicing philosophers! More seriously, Blackwell Publishers Ltd. 2002 220 G. Hatﬁeld Wilson’s claim that philosophers felt threatened by psychologists is belied by
data in an article he cites. Ruckmich (1912), comparing the standing of psychology in 1911 to philosophy and other ﬁelds in 39 institutions, found that
in nearly every category of comparison, whether total budgeted faculty salaries,
number of students taught, or number of students matriculated, philosophy
ranked above psychology in most universities. For the nineteen schools that
released budgetary information, the average total budget for salaries was $6545
for philosophy, and $5285 for psychology (compared with $15 545 for physics,
and $11 090 for zoology). On these measures, philosophy and psychology both
ranked below some sciences, and also below some departments of education,
revealing the American penchant for practicality and application.
The actual relations between philosophy and psychology in American universities were reasonably amiable. The two disciplines typically were not set
in competition for resources, and both experienced tremendous growth during
the 1890s and into the ﬁrst decades of the twentieth century. The situation was
totally different from that in Germany. There had been no graduate schools in
Arts and Sciences in America until late in the century. During the course of
the century, and especially after the Civil War, instruction in science had
increased at major schools, including Harvard College with its Lawrence Scientiﬁc School and the University of Pennsylvania with its Towne School. By
contrast, both philosophy and psychology were typically taught by a man with
religious credentials (often a reverend), who served as Provost or Vice Provost
of the college or university. When doctoral programs in Arts and Sciences
came to America in the 1880s and 1890s, many ﬁelds underwent an expansion
of faculty (see Geiger, 1986; Veysey, 1965). Philosophy and psychology started
with comparatively few faculty members (often the same person). But in the
general expansion, both were allowed to grow. Equipment was cheap in both
ﬁelds (cheaper in philosophy). Already in the 1890s separate positions were
established in experimental psychology, and as universities organized into more
ﬁne-grained departmental structures, philosophy and psychology tended to go
their separate ways. Of the 39 psychology programs surveyed by Ruckmich
(1912), 21 had completely distinct departments, and in the remaining cases,
the afﬁliation with philosophy (or sometimes with education) was frequently
characterized as ‘partial’ or ‘theoretical’ (i.e., notional), and in about half the
cases the psychologists reported that they did not believe the afﬁliation posed
a problem (data was not collected by Ruckmich on how the philosophers felt).
Overall, there was no need for war, for both ﬁelds were experiencing a rapid
expansion of resources.
A genuine spirit of intellectual interchange pervades the discussions of
behaviorism by philosophers and psychologists in the teens and twenties.
Further interchange is evident from the Minnesota Studies in the Philosophy
of Science volumes in the 1950s, and from books such as Hamlyn (1957).
Philosophy and psychology remain in dialogue. And with the more recent rise
of cognitive science, an especially close relationship obtains between philo Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 221 sophy and other disciplines that study cognition; indeed, a philosopher (Fodor,
1975) offered the core statement of the assumptions that provided theoretical
unity to cognitive science itself. Questions remain, however, about the relation
between psychology and cognitive science.
4. Psychology and Cognitive Science
There is a general impression that during the period from 1920 to the mid
1950s, behaviorism succeeded in driving cognitive topics and cognitive theoretical notions out of experimental psychology in America. This impression
feeds a story that as a result of symposia on neuroscience, artiﬁcial intelligence,
and information theory in the period 1948 to 1956, cognitive science was
born, which in turn allowed cognitive psychology to develop (Gardner 1985).
Allegedly, in the meantime Chomsky (1959) ﬁnished off behaviorism with
his review of Skinner’s Verbal Behavior. Cognitive science, as a mixture of
computational and linguistic models (with some hope for a connection with
neuroscience), was here to stay.
In fact, each of these assertions should be questioned—and, in my view,
rejected. Although behaviorism became strong or even dominant in the period
1920–1960, it by no means was able to stamp out the study of cognition and
perception in American psychology. For one thing, the Gestalt psychologists
W. Kohler and K. Koffka—open opponents of behaviorism who studied
thought and perception using phenomenological methods—immigrated during
the 1930s and established the Gestalt viewpoint as one to be reckoned with.
Further, investigators originally trained in the Gestalt tradition, such as Irvin
Rock (1954) and William Epstein (1973), converted to a view of perceptual
processing as a combining, through cognitive or non-cognitive processes, of
information registered (perhaps pre-consciously) by the perceptual system.
But even beyond the ongoing work in perception, there remained a tradition of studying attention, memory, problem solving, and thought. Woodworth (1938), a popular handbook in experimental psychology, contained
chapters discussing such behaviorist favorites as conditioning and maze learning. But it had many more chapters on perceptual and cognitive topics, including the use of reaction time to measure the ‘time of mental processes’ (ch.
14), and separate chapters on attention, reading, problem solving, and thinking.
The latter chapter had a discussion of ‘anticipatory schema’ in thinking, and
of ‘frames’ as ‘a performance in outline, needing to be ﬁlled in’ (1938, p. 798).
If anything, the cognitive content was only increased in Osgood’s Method and
Theory in Experimental Psychology (1953), which in its discussions of learning,
problem solving, and thinking introduced what was termed Tolman’s ‘cognition theory’, and freely discussed positing, to explain both animal and human
performance, ‘representational mediation processes’, described as ‘symbolic
processes’ (1953, pp. 382, 401, 663).
It is true that the rise of the computer, and the discussions of information Blackwell Publishers Ltd. 2002 222 G. Hatﬁeld processing in the 1950s, provided new models and analogies psychologists
could use in framing theories of problem solving or other perceptual or cognitive tasks. The analogy of the computer was cited explicitly as providing
grounds for thinking that the internal workings of complex information-handling devices could be understood at the ‘program level’, and could be
instantiated in a physical device whose operation was theoretically explicable
(Green, 1967). This comparison helped some psychologists overcome the
behaviorist aversion to positing internal processes—though as we just saw,
Osgood had already found the means for that in the work of Tolman and
In fact, early computer analogies in psychology did not yield computational
theories of the sort described by Fodor (1975), and characteristic of cognitive
science during the 1980s. Neisser’s Cognitive Psychology (1967) is a case in point.
It adopted a ﬂow-chart model for tracking information through various processes by which it was received and transformed, and it compared the psychologist’s task to that of discovering the ‘program’ for human cognition. But it
expressed skepticism concerning treating programs as models of actual psychological processes in which the computational steps in a computer are compared
to fundamental psychological processes themselves. As Neisser put it, ‘the rest
of the book can be considered as an extensive argument against models of this
kind’ (1967, p. 9). And in fact his book showed no sign of adopting a computational model of mind such as that described by Fodor (1975). My own
impression is that the majority of American psychologists studying perception
and cognition did not and have not fully joined the ‘cognitive science’ movement. In the ﬁnal section I explain this as a result of their distinctive approach
to experimentation and theorizing about perception and cognition.
Cognitive science itself, in its ‘classical’ formulation, was a mixture of artiﬁcial intelligence and linguistics, drawing on some areas of cognitive psychology, and sometimes attempting to make connections with neuroscience.
Much of the history of this formulation has been told by Gardner (1985),
though his chapter on psychology contains comparatively little on the recent
contributions of psychology to cognitive science, focusing mainly on historical
theories of perception and cognition, such as those of Wundt and the Gestaltists
(1985, ch. 5). The theoretical center piece of traditional cognitive science is the
computer analogy, or, more accurately, the assertion that animal and human
psychology occurs through computational processes in organisms involving
physical symbol systems (Fodor, 1975; Newell, 1980). The idea is that psychological processes such as those studied in cognitive and perceptual psychology
are realized, in animals and humans, by ‘sub-personal’ (hence pre-conscious)
innate symbol systems in which the programs of the mind are initially given
and are subsequently modiﬁed (through experience). This conception offers a
theory of how cognition really takes place in the organism: through the interaction of symbols that are processed by internal operations sensitive to their
syntactic structure. In the end, this view holds that the mind actually runs like Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 223 a computer. In the most updated versions, this computer is conceived as a
massively parallel, modularized, naturally evolved device, but as a symbol-using
computational machine nonetheless (Pinker, 1997).
A prominent case of a psychologist adopting this sort of computer analogy
was Stephen Kosslyn’s (1980, 1983) original model of visual imagery. In this
model Kosslyn developed a Neisser-like decomposition of human imaging
capacities, ﬁtted together into a ﬂow chart of processing activities. He then
imagined these operations to be realized in digitally encoded data arrays (with
color and lightness values assigned to ‘cells’ organized into columns and rows,
accessed through logical addresses, and not as a physically real matrix of spatially
contiguous locations). This theory ﬁt so well the standard model of what a
‘classical’ computational theory should be, while also engaging the experimental data of psychology, that Barbara von Eckardt (1993) took it as a paradigm for work in cognitive science.
The problem is that the computer-analogy part of Kosslyn’s original theory
was like a free wheel spinning. The detailed ﬂow charts and functional
decomposition of imagery tasks into such operations as ‘lookfor’, ‘scan’, and
‘put’ did real work and could be made to face the tribunal of experimental
conﬁrmation, e.g., by studying the reaction times of subjects who performed
imaging tasks. The hypothesis that these operations are instantiated in an
underlying symbol system added nothing essential to the theory of imagery.
It had the virtue of offering a precise characterization of how the operations
are carried out, but as Anderson (1978) observed, any psychological process
might be modeled by brute force using symbolic processing. Kosslyn at ﬁrst
considered dropping symbol systems (Kosslyn and Hatﬁeld, 1984). He subsequently (1994) altered his conception of the realization of images, to spatially
organized arrays of ‘points’, neurally instantiated in analog form (1994, pp.
12–20). Curiously, he continued to call the points composing his depictive
representations ‘symbols’ (1994, pp. 280–2). That label is misleading, since in
his new model no operations are deﬁned which respond to these points based
on variation in their individual forms (as opposed to the depictive forms they
collectively instantiate), as happens in classical symbol-processing models
(Fodor, 1975, ch. 2; Pylyshyn, 1984, ch. 3).
Kosslyn’s (1980) was a bold attempt to bring the new computational and
symbolic theories of cognitive science into experimental psychology. Leaving
aside his retention of the bare term ‘symbol’, Kosslyn has moved away from
the classical conception. Many of his colleagues in psychology never went
there, or only ﬂirted with the view and have now moved away.
Larry Barsalou (1999) has developed a line of thinking which promises to
synthesize the perspectives of cognitive science with those of experimental
psychology. However, unlike the early work of Kosslyn, which grafted a symbolist undercarriage onto a theory of imagery that could in fact stand on its
own (or be joined to work in neuroscience), Barsalou is using psychological
ﬁndings to criticize assumptions in classical cognitive science. He is turning Blackwell Publishers Ltd. 2002 224 G. Hatﬁeld the direction of purported inﬂuence around, and suggesting that cognitive
science has proceeded too far without adequate input from perceptual and
cognitive psychology. His target is in particular the ‘amodal’ (i.e., represented
in a manner independent of the various sensory modalities such as vision or
hearing) symbols of Fodor (1975), Pylyshyn (1984), and others. In other words,
he is going after the core of classical symbolist cognitive science.
Barsalou’s point is that much of cognition relies on speciﬁcally perceptual
modes of representation. He contends that concepts are typically stored in
forms that represent the way things look, sound, smell, feel, etc., in perceptual
(e.g., analog, as opposed to amodal symbolic) form. His argument draws on
a large body of empirical work in psychology on memory and concepts. Some
of this work tests the limitation of amodal theories by showing that concepts
typically have a perceptual character. Work with neurological deﬁcits reveals
a connection between sensorimotor areas of the brain and conceptual abilities.
Barsalou does not, of course, deny that humans have amodal symbolic abilities,
or that there is a psychology of such abilities. What he denies is the classical
conception that amodal symbols are the coin of the realm of the psychological
processes underlying perception and cognition, as classical cognitive science
would have it (see also Hatﬁeld 1988, 1991).
The classical conception in cognitive science is under attack from many
fronts, including the connectionist alternative to amodal symbolic processing.
Connectionist models can themselves be conceived as realizing either amodal
or perception-based representations. With the growth of sub-symbolic connectionist models (Smolensky, 1988), the notion that cognitive science is
deﬁned by symbolic computational processes is fading. But there is room for
an alternative conception of cognitive science, one that does not demand that
theorists use symbol-processing operations to instantiate all processing models.
The mode of instantiation would remain open. Investigation could then proceed along the traditional lines of functionally decomposing psychological
mechanisms, without yet adopting any particular micro-model of how psychological capacities are realized in the brain. Having dropped the imperialist
demand for symbol-processing, cognitive science could now become a confederation of independent disciplines and perspectives—including psychology,
philosophy, linguistics, computer science, and sometimes anthropology and
neuroscience—each with its own methods and/or theories to contribute to
the mix. This federational view of cognitive science has much to recommend
over the symbolist-imperialist vision of Fodor and Pylyshyn. The proof of this
brand of cognitive science is of course whether the various disciplines do have
their own distinctive contributions to make. Focusing as we are on psychology,
that brings us to our ﬁnal question.
5. Will Psychology Go Away?
Gardner (1985) foresees that cognitive psychology will join forces with artiﬁcial
intelligence and be absorbed into cognitive science. Gazzaniga (1998) contends Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 225 that any part of psychology not going in that direction should be absorbed
by neuroscience. By contrast, I think that psychology is here to stay for the
Ernest Hilgard, a distinguished psychologist and historian of psychology
with personal experience of much of twentieth-century American psychology
(he was born in 1904), recently considered the question of what keeps psychology together as a uniﬁed discipline. He acknowledged the centrifugal
forces that draw psychology into contact with other disciplines, including
linguistics, artiﬁcial intelligence, biology, neuroscience, anthropology, and
sociology. Yet, psychology retains its own identity. He explained:
What binds us together are agreement upon a preference for experimental
approaches, the use of appropriate statistics in determining the reliability
of such ﬁndings, and a preference for theories that integrate such ﬁndings.
We have attained status as a legitimate social science and also a biological
science, depending on the subﬁelds under consideration. While we may
expect changes, our role as a legitimate member of the scientiﬁc disciplines
appears to be assured (1997, p. xv).
Experiment and experimental design, characteristic use of statistical analysis,
and development of theories that integrate empirical data gained through
experiment. These are core values that distinguish psychological research from
the sort of data collection and theorizing that go on in neuroscience, artiﬁcial
intelligence, linguistics, and anthropology. Psychologists may indeed enter into
cooperative arrangements with these other ﬁelds—testing a computational
model against data, specifying or assessing normal and pathological psychological function for the neurologist, proposing psychological functions so that the
neuroscientist can look for their neural realization. But it seems unlikely that
their distinctive contribution will be replicated in these other ﬁelds, without
looking to psychologically trained specialists.
An appreciation of the basis of my prediction that psychology will retain
its distinctive role can be gained by comparing three recent books in the interdisciplinary ﬁeld of vision science. The books are Brian Wandell (1995), who
integrates mathematical modeling with brain science, Hanspeter Mallot (2000),
who takes a computer vision approach, and Stephen Palmer, who adopts an
information processing approach, as seen from ‘a psychological perspective’
(1999, p. xix). All three are good books, and they draw on some of the same
literature. But the ﬁrst two differ from the third in the nearly total lack of
psychological models of perceptual processing, and nearly total lack of citation
to psychological experiments. Let’s see what this difference amounts to.
All three books discuss in detail the optical information available at the
retina and the question of what is needed to extract information about the
environment from that information. They all see the information available as
in some respects ambiguous or underdetermining of the distal scene. This Blackwell Publishers Ltd. 2002 226 G. Hatﬁeld means that cognitive or psychological operations must be posited to process
the information so as to yield typically accurate perception and cognition of
distal scenes. (We do typically succeed in discovering by vision many properties
of surrounding objects.) All three invoke the notion of ‘information processing’
in characterizing these operations. Mallot suggests that the observer draws on
‘prior knowledge or assumptions about the environment’ to infer back from
the image to the environment (2000, p. 14). Wandell proposes that the visual
system uses statistical regularities to interpret the environment by drawing
‘accurate inferences about the physical cause of the image’ (1995, p. 3). Palmer
also sees the problem as one of making appropriate inferences on the information given.
The three books differ greatly in whether and how they characterize and
discuss animal and human information processing. Mallot makes comparatively
little reference to the literature of neuroscience or psychology, and includes
almost no discussion of neurophysiology, postulated psychological processes,
or behavioral tests of living systems. That is not the aim of his book. Rather,
he has written a book that takes a ‘computational theory’ approach (in the
sense of that term deﬁned in Marr, 1982). As Mallot puts it, ‘the central issue
is of what must be calculated in order to derive the desired information from
the spatio-temporal stimulus distribution. . . On the level of computational
theory, the brain and the computer are confronted with the same problems’
(2000, p. 6). In other words, he is not interested in the problem of how
biological perceivers actually carry out the information processing needed in
order to perceive. Rather, he is interested in characterizing what mathematical
relations hold between environment and senses, and what plausible assumptions about the structure of the environment would allow the recovery of
information about the distal scene from the optical stimulus. As he puts it, this
simply is ‘the problem of computer vision’ (2000, p. 8). His various chapters
then show in some detail how a mathematical reconstruction of the distal scene
can be achieved using plausible assumptions. Such work provides highly useful
information for building robots that can do visual tasks, or for characterizing
tasks that might be carried out by biological systems. But it clearly does not
directly engage the psychology of vision in living systems.
Wandell’s basic approach shares Mallot’s emphasis on computational analysis, but adds data from neuroscience. Mallot did not discuss even basic neuroanatomy or physiology. By contrast, Wandell covers the photoreceptors, neural
pathways, and cortical physiology (more extensively than Marr, 1982, who
earlier sought to wed the computational approach with neural
implementation). He distinguishes the job of ‘learning how we might see’ by
studying the inference rules needed to recover information from stimulation,
from the task of using ‘neural and behavioral measurements of our brains and
our perceptions’ in order to ‘learn how we do see’ (1995, p. 402). He has an
interest in neural and behavioral reality. At the same time, he admits that in
our current state of knowledge we know more about the description of stimu Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 227 lation and photoreceptor response (the ‘encoding’) and the immediately subsequent neurophysiological representation, than we do about the ‘interpretation’ stage that governs how we see. Consequently, in Part 3 of his book,
on interpretation, he presents mainly three types of information: ﬁrst, the sort
of computational possibilities for interpretation emphasized by Mallot, to
which he adds a more extensive discussion of linear models in color constancy;
second, neurophysiological studies of the pathways and localization of the neural activity in vision; and third, evidence that certain neurons are sensitive to
the sorts of information that his mathematical models describe in stimuli. But
as Wandell himself contends, to understand how vision works we need not
so much to know where the neural activity is, but how it serves vision (1995,
pp. 191, 336). The discussion of front-end edge or motion detectors ties his
mathematical models to biological reality. But in preparing to discuss the interpretive, inferential processes he believes underlie vision, he concludes that the
‘neuron doctrine’ of looking for ever more comprehensive single-cell ‘detectors’ won’t work, and that further behavioral and computational studies are
needed (1995, pp. 188–91). These studies will aim at correlating discriminative
capacities with manipulations of the type of information described in his computational models. This would make for an elaborated psychophysics (and as
such would bear resemblance to the psychologist James J. Gibson’s theories of
vision, 1950, 1966), but it leaves out the entire body of psychological literature
that attempts to experimentally analyze the processing mechanisms themselves,
by relating performance data, including reaction time, error rates, and
phenomenal report, to cleverly constructed perceptual tasks. Wandell’s ﬁnal
chapter on ‘seeing’ is an invitation to try to discover how we see. The only
psychological theory discussed is Gregory’s (1966) theory of the Muller-Lyer
illusion, along with an experimental counter-example to its predictions.
Palmer’s (1999) book is written as a contribution to the interdisciplinary
ﬁeld of vision science, considered as a subﬁeld of cognitive science. It attempts
to integrate ﬁndings from computer models, neuroscience, and experimental
psychology. It differs from Wandell’s book in three important respects. First,
although Wandell’s book examines computational theories in some detail, it
generally avoids discussion of theories of organismic processes of information
processing that result in vision itself (as opposed to those that code retinal
information). Palmer devotes an entire chapter to such theories, including the
classical theories of Helmholtz, the Gestaltists, and Gibson, computer theories,
biological information processing theories, Marr’s hybrid computational and
biological theory, and psychological information processing theories. He also
carries the discussion of the various theoretical perspectives to each of the
subsequent subject areas, including color vision, spatial vision, motion and
events, and visual awareness. Second, although Wandell mentions visual awareness, he has little to say about it. Palmer takes visual experience itself to be
an important object of explanation in vision science. Third, Palmer exhaustively surveys the literature in perceptual psychology that bears on the problems Blackwell Publishers Ltd. 2002 228 G. Hatﬁeld of spatial, color, and motion perception. This means he includes theoretical
issues that Wandell and Mallot, with their computational and neural focus,
leave out (and only some of which Marr approached, with scant reference to
the psychological literature). These include unconscious inference as opposed
to relational theories of lightness constancy, experiments and theories on perceptual organization, theories and experiments on a variety of spatial constancies, and experimental and theoretical work on motion perception, apparent motion, and grouping by movement. These various lines of work attempt
to analyze the actual psychological processes underlying perception.
In the end, Palmer contends that interdisciplinary vision science has advantages over isolated single-discipline work, because it combines the perspectives
of the independent disciplines and recognizes that each has something to contribute. As he puts it:
Vision science may have made the boundaries between disciplines more
transparent, but it has not eliminated them. Psychologists still perform
experiments on sighted organisms, computer scientists still write programs
that extract and transform optical information, and neuroscientists still
study the structure and function of the nervous system (1999, p. xix).
The use of experiments on organisms not simply to determine discriminative
capacities and psychophysical correspondences, but to dissect internal processes,
remains a fundamental contribution of the psychological study of vision.
Of course, one might believe that visual scientists can converge on these
processes working from computations, discriminative capacity, and
neuroscientiﬁc measurements alone. Wandell seems to think so. And yet
neuroscientiﬁc treatments (e.g., as surveyed in Kandel, Schwartz, and Jessell,
2000), like Wandell’s book, lack discussion of theories of information processing, beyond the initial detector-like registration of features, or investigations of gross anatomical location. Moreover, it seems that to discover the
neural instantiation of visual processes, one will need a characterization not
only of what the processes accomplish (input-output relations), but of how it
is accomplished (functional organization of the processes). At present, such
functional level processing in biologically real systems is primarily the responsibility of psychological investigations. But since one needs a functional level
theory in order to ask whether and how the brain realizes the processes
described in the theory, it seems that brain science will remain dependent
on psychological science for characterizations of global brain functioning in
perception (Hatﬁeld, 2000).
There is, of course, no real assurance that the discipline of psychology will
retain its identity. Gazzaniga reports that his dean once told him that if the
Psychology Department would rename itself the Department of Brain and
Cognitive Science, ‘I could raise $25 million in a week’ (1998, p. xii). That
sort of consideration might settle the issue. But if the newly named department Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 229 is to succeed in the study of perception and cognition, it will need to use the
theories and methods of psychology. The study of those brain functions of
interest to cognitive science is the study of the psychological processes of
organisms. The structure of those processes is not read off single cell recordings
or images of brain activity. Rather, those recordings and images gain meaning
by being related to a theory of what’s being done functionally. Psychology is
here to stay, even if disguised so as to fool the money lenders. There’s no way
Department of Philosophy
University of Pennsylvania References
Amundson, R. 1983: E.C. Tolman and the intervening variable: A study in the epistemological history of psychology. Philosophy of Science, 50, 268–82.
Amundson, R. 1986: The unknown epistemology of E.C. Tolman. British Journal of
Psychology, 77, 525–31.
Amundson, R. and Smith, L.D. 1984: Clark Hull, Robert Cummins, and functional
analysis. Philosophy of Science, 51, 657–66.
Anderson, J.R. 1978: Arguments concerning representations for mental imagery.
Psychological Review, 85, 249–77.
Barsalou, L. 1999: Perceptual symbol systems. Behavioral and Brain Science, 22, 577–660.
Bergmann, G., and Spence, K.W. 1941: Operationism and theory in psychology.
Psychological Review, 48, 1–14.
Bonnet, C. 1755: Essai de psychologie; ou, considerations sur les operations de l’ame, sur
l’habitude et sur l’education. London.
Boring, E.G. 1929: History of Experimental Psychology. New York: Century.
Boring, E.G. 1950: History of Experimental Psychology, 2d edn. New York: Century.
Bridgman, P.W. 1927: Logic of Modern Physics. New York: Macmillan.
Burtt, E.A. 1925: Metaphysical Foundations of Modern Physical Science: A Historical and
Critical Essay. London: Kegan Paul.
Carnap, R. 1932: Psychologie in physikalischer Sprache. Erkenntnis, 3, 107–42. Trans.
1959 G. Schick, Psychology in physical language. In A.J. Ayer (ed.), Logical
Positivism. New York: Free Press, 165–97.
Chomsky, N. 1959: Review of B.F. Skinner’s Verbal Behavior. Language, 35, 26–58.
D’Arcy, P. 1765: Memoire sur la duree de la sensation de la vue. Memoires de mathema´
tique et de physique, 82 (Paris edition), 439–51.
Danziger, K. 1980: The history of introspection reconsidered. Journal of the History of
the Behavioral Sciences, 16, 241–62.
Epstein, W. 1973: The process of ‘taking into account’ in visual perception. Perception,
11, 75–83. Blackwell Publishers Ltd. 2002 230 G. Hatﬁeld Fodor, J.A. 1975: Language of Thought. New York: Crowell.
Gardner, H. 1985: The Mind’s New Science: A History of the Cognitive Revolution. New
York: Basic Books.
Gazzaniga, M.S. 1998: The Mind’s Past. Berkeley, Calif.: University of California Press.
Geiger, R.L. 1986: To Advance Knowledge: The Growth of American Research Universities,
1900–40. Oxford University Press.
Gibson, J.J. 1950: Perception of the Visual World. Boston: Houghton Mifﬂin.
Gibson, J.J. 1966: Senses Considered as Perceptual Systems. Boston: Houghton Mifﬂin.
Godart, Guillaume-Lambert. 1755: La physique de l’ame. Berlin.
Green, B.F. 1967: The computer conquest of psychology. Psychology Today 1 (no. 7),
Gregory, R.L. 1966: Eye and Brain: The Psychology of Seeing. London: Weidenfeld
Hamlyn, D.W. 1957: The Psychology of Perception: A Philosophical Examination of Gestalt
Theory and Derivative Theories of Perception. London: Routledge.
Hatﬁeld, G. 1988: Representation and content in some (actual) theories of perception.
Studies in History and Philosophy of Science, 19, 175–214.
Hatﬁeld, G. 1990a: Metaphysics and the new science. In D. Lindberg and R. Westman
(eds.), Reappraisals of the Scientiﬁc Revolution. Cambridge University Press, 93–166.
Hatﬁeld, G. 1990b: The Natural and the Normative: Theories of Spatial Perception from
Kant to Helmholtz. Cambridge, Mass.: MIT Press.
Hatﬁeld, G. 1991: Representation and rule-instantiation in connectionist systems. In
T. Horgan and J. Tienson (eds.), Connectionism and the Philosophy of Mind. Boston:
Hatﬁeld, G. 1995: Remaking the science of mind: Psychology as a natural science. In
C. Fox, R. Porter, and R. Wokler (eds.), Inventing Human Science. Berkeley: University of California Press, 184–231.
Hatﬁeld, G. 1996: Was the scientiﬁc revolution really a revolution in science? In Jamil
Ragep and Sally Ragep (eds.), Tradition, Transmission, Transformation. Leiden: Brill,
Hatﬁeld, G. 1997: Wundt and psychology as science: Disciplinary transformations.
Perspectives on Science, 5, 349–82.
Hatﬁeld, G. 2000: The brain’s ‘new’ science: Psychology, neurophysiology, and constraint. Philosophy of Science, 67, S388–403.
Hatﬁeld, G. Forthcoming: Perception as unconscious inference. In D. Heyer and R.
Mausfeld (eds.), Perception and the Physical World. Oxford University Press.
Hempel, C.G. 1935: Analyse logique de la psychologie. Revue de synthese, 10, 27–42.
Trans. 1949 W. Sellars, Logical analysis of psychology. In H. Feigl and W. Sellars
(eds.), Readings in Philosophical Analysis. New York: Appleton-Century-Crofts,
Hilgard, E.R. 1987: Psychology in America: A Historical Survey. San Diego: Harcourt
Hilgard, E.R. 1997: A personal view of 20th-century psychology: With an eye to the
21st century. In R.L. Solso (ed.), Mind and Brain Sciences in the 21st Century.
Cambridge, Mass.: MIT Press, xi–xv. Blackwell Publishers Ltd. 2002 Psychology, Philosophy, and Cognitive Science 231 Hull, C.L. 1943: Principles of Behavior: An Introduction to Behavior Theory. New York:
James, W. 1890: Principles of Psychology, 2 vols. New York: Holt.
Kandel, E.R., Schwartz, J.H., and Jessell, T.M. 2000: Principles of Neural Science, 4th
edn. New York: McGraw-Hill.
Koch S. 1941: The logical character of the motivational concept. Psychological Review,
Koch S. 1964: Psychology and emerging conceptions of knowledge as unitary. In T.W.
Wann (ed.), Behaviorism and Phenomenology: Contrasting Bases for Modern Psychology.
University of Chicago Press, 1–41.
Kohler, W. 1929: Gestalt Psychology. New York: Liveright.
Koffka, K. 1935: Principles of Gestalt Psychology. New York: Harcourt, Brace.
Kosslyn, S.M. 1980: Image and Mind. Cambridge, Mass.: Harvard University Press.
Kosslyn, S.M. 1983: Ghosts in the Mind’s Machine: Creating and Using Images in the Brain.
New York: Norton.
Kosslyn, S.M. 1994: Image and Brain: The Resolution of the Imagery Debate. Cambridge,
Mass.: MIT Press.
Kosslyn, S.M., and Hatﬁeld, G. 1984: Representation without symbol systems. Social
Research, 51, 1019–45.
Kruger, Johann Gottlob. 1756: Versuch einer Experimental-Seelenlehre. Halle.
Kusch, M. 1995: Psychologism: A Case Study in the Sociology of Philosophical Knowledge.
Leahey, T.H. 1980: A History of Psychology. Englewood Cliffs, N. J.: Prentice-Hall.
Loeb, J. 1900: Comparative Physiology of the Brain and Comparative Psychology. New
York: G.P. Putnam and Sons.
Mallot, H.A. 2000: Computational Vision: Information Processing in Perception and Visual
Behavior. Cambridge: MIT Press.
Marr, D. 1982: Vision. San Francisco: W. H. Freeman.
McDougall, W. 1905: Physiological Psychology. London: Dent.
McDougall, W. 1912: Psychology, The Study of Behaviour. New York: Holt.
Neisser, Ulric. 1967: Cognitive Psychology. New York: Appleton-Century-Crofts.
Newell, A. 1980: Physical symbol systems. Cognitive Science, 4, 135–83.
O’Donnell, J.M. 1979: The crisis in experimentalism in the 1920s: E.G. Boring and
his uses of history. American Psychologist, 34, 289–95.
Osgood, Charles Egerton. 1953: Method and Theory in Experimental Psychology. New
York: Oxford University Press.
Palmer, S.E. 1999: Vision Science: Photons to Phenomenology. Cambridge: MIT Press.
Pillsbury, W.B. 1911: Essentials of Psychology. New York: Macmillan.
Pinker, S. 1997: How the Mind Works. New York: Norton.
Pylyshyn, Z. 1984: Computation and Cognition: Toward a Foundation for Cognitive Science.
Cambridge: MIT Press.
Roback, A.A. 1923: Behaviorism and Psychology. Cambridge, Mass.: University Book
Store. Blackwell Publishers Ltd. 2002 232 G. Hatﬁeld Rock, I. 1954: The perception of the egocentric orientation of a line. Journal of Experimental Psychology, 48, 367–74.
Ruckmich, C.A. 1912: History and status of psychology in the United States. American
Journal of Psychology, 23, 517–31.
Schmid, Carl C.E. 1796: Empirische Psychologie, 2d ed. Jena: Croker.
Singer, E.A. 1911: Mind as an observable object. Journal of Philosophy, Psychology and
Scientiﬁc Methods, 8, 180–86.
Skinner, B.F. 1938: Behavior of Organisms: An Experimental Analysis. New York:
Skinner, B.F. 1945: Operational analysis of psychological terms. Psychological Review,
52, 270–77, 291–4.
Skinner, B.F. 1957: Verbal Behavior. New York: Appleton-Century-Crofts.
Smith, L.D. 1981: Psychology and philosophy: Toward a realignment, 1905–1935.
Journal of the History of the Behavioral Sciences, 17, 28–37.
Smith, L.D. 1986: Behaviorism and Logical Positivism: A Reassessment of the Alliance. Stanford: Stanford University Press.
Smolensky, P. 1988: On the proper treatment of connectionism. Behavioral and Brain
Sciences, 11, 1–74.
Titchener, E.B. 1912: The schema of introspection. American Journal of Psychology, 23,
Tolman, E.C. 1926: A behavioristic theory of ideas. Psychological Review, 33, 352–69.
Tolman, E.C. 1927: A behaviorist’s deﬁnition of consciousness. Psychological Review,
Tolman, E.C. 1932: Purposive Behavior in Animals and Men. New York: Century.
Tolman, E.C. 1951: Collected Papers in Psychology. Berkeley: University of California
Turner, R.S. 1994: In the Eye’s Mind: Vision and the Helmholtz-Hering Controversy.
Princeton, New Jersey: Princeton University Press.
University of Pennsylvania. 1938: The college. Bulletin, 38 (no. 24).
Veysey, L.R. 1965: Emergence of the American University. University of Chicago Press.
Von Eckardt, B. 1993: What Is Cognitive Science?. Cambridge, Mass.: MIT Press.
Wandell, B. 1995: Foundations of Vision. Sunderland, Mass: Sinauer.
Watson, J.B. 1913: Psychology as the behaviorist views it. Psychological Review, 20,
Watson, J.B. 1914: Behavior: An Introduction to Comparative Psychology. New York: Holt.
Watson, J.B. 1919: Psychology from the Standpoint of a Behaviorist. Philadelphia: Lippincott.
Watson, J.B. 1925: Behaviorism. New York: Norton.
Wilson, D.J. 1990: Science, Community, and the Transformation of American Philosophy,
1860–1930. University of Chicago Press.
Woodger, J.H. 1938: The Axiomatic Method in Biology. Cambridge University Press.
Woodworth, R.S. 1938: Experimental Psychology. New York: Holt.
Wundt, W. 1902: Outlines of Psychology, tr. C. H. Judd. Leipzig: Engelmann. Blackwell Publishers Ltd. 2002 ...
View Full Document
This note was uploaded on 12/09/2011 for the course SP 108 taught by Professor Whittenburg during the Summer '11 term at Montgomery College.
- Summer '11
- Main Street