Summary and Review for Experience and Content: Consequences of a Continuum Theory

W. Martin Davies; Hants: Ashgate Press, 1996;
Avebury Series in Philosophy; ISBN: 1-85972-342-X (HB)

SUMMARY

This book was my first PhD thesis. It is broadly concerned with the category of observational experience in philosophical thought and the influence of recent ‘propositional linguistic’ views on theories of mind and content.

Kant is long credited with arguing that experience is importantly structured by epistemic categories and, generally, ‘high-level’ organisation. Taken together with the ‘theory dependence of observation’ thesis in the philosophy of science, this view has become something akin to dogma. I argue that the stress on observational experience being ‘inferential’, theory dependent and crucially underpinned by linguistic and epistemic categories is misplaced, and that there is some sense in which experience is non-inferential and sensory in content. I argue by implication that the ‘inferentialist’ views of Feyerabend, Churchland, Kant, Armstrong and Sellars is misplaced and that content can be understood in more subtle terms.  The theory/observation threshold does not collapse with high-level influences necessarily taking the dominant organisational role.

In the thesis I draw upon the work of Fodor, Nagel, Peacocke, Millar, Jackson and Dretske to legitimate a somewhat less extreme position. By directly confronting the arguments of ‘inferentialist’ thinkers, and including for consideration the issue of animal and infant cognition and problems in perceptual psychology, I advance the view that experiential content may have a number of non-exclusive elements-a low-level, yet structured, sensory content being one such element.  By drawing upon a phylogenetic argument and what I call a ‘continuum’ theory of content, I suggest that current views of experience are too heavily inferential in emphasis to be plausible-a ‘dual-aspect’ Nagelian-type position is a consequence of understanding content in this way. Hence, I repudiate the inference laden linguistic views currently popular today and opt for an account which allows structure at several non-exclusive levels.

The thesis is divided into four parts: a general section, a section on the structure of mental content and sections on experience as a category in the philosophy of mind and the philosophy of science.

Reviews

Experience and Content has been reviewed in the following places:

  • David Lumsden, Australasian Journal of Philosophy Vol 76, No. 3, September (1998): pp. 508-510
  • Vladimir Popescu, Metascience, June 1998, pp. 144-148
  • Andrew Latus, Philosophy in Review (Canadian Review of Philosophy), Vol. 18, No. 2, 1998, pp. 92-94.
  • Dennis Lomas, Psyche, Vol 6, No. 7, 2000.

Contents

Acknowledgements
Preface
Introduction

Part I: Experience and Content

1.     Observations and Inferences
2.     A Continuum Theory of Content
3.     Experience and Language
4.     Experience and Beliefs

Part II: Experience and Structure

5.    Sensational Experiences
6.    Idealised and Naturalised Experience
7.    Animal Experiences
8.    Modularity and Insularity

Part III: Experience and Science

9.    Experience Without Feyerabend

Part IV: Experience and Mind

10.    Sellars’ Myth
11.    Experience Eliminated?
12.    Experience and Subjectivity

Conclusion: Experiencing the “Manifest Image”

Bibliography

Back to top

REVIEWS

Review 1: Dr David Lumsden

Davies addresses the issue of whether experiential content is inferential, in the sense that content depends on some or all of various ‘high level features’ including background knowledge, existing concepts, theories or language. Different varieties within this inferentialist camp are distinguished and the term covers the views of Descartes, Kant, Sellars, Armstrong and the Churchlands, amongst others.

Davies argues against all varieties of the inferentialist view but does not adopt a purely non-inferential view such as a sense datum theory. He argues for a compromise view which admits the existence of non-inferential content, such as qualia, but only at a certain level. Taking his cue from Fodor’s modular view in which inference only kicks in at a certain level and, following a brief suggestion by Armstrong, he argues for his continuum hypothesis, which involves a multi-layered approach, in which we can distinguish between different degrees of high level features in experience.

The notion is that at the bottom end there are no high level features, just purely sensational experience, and as we ascend to higher levels there are greater and greater admixtures of high level features, with the higher level always including all that came before. The term ‘continuum’ might seem out of place as the discussion appears at times to involve a limited number of discrete levels of experience, but is supported by the notion that high level features can be present to varying degrees. There is a diagram of the levels on p. 48 which puzzled this reader, as it appeared to be a mix of a one dimensional and a two dimensional representation.

This main theme of the book, as described above, is introduced in the first two chapters. The third chapter of Part 1, Experience and Content, is a chapter on experience and language in which it is argued that it is a serious error to move from the view that observational terms are theory laden to the view that experience is always theory laden. The fourth chapter is on experience and belief. The issue of whether perception involves the acquisition of beliefs is addressed using the continuum approach. A continuum of levels of belief is embraced, from ones with sophisticated representational characteristics down to different kinds of look-beliefs which are determined by sensational features of perception and do not involve high-level features.

Part 2, Experience and Structure, (chapters five to eight) develops themes already mentioned and is arguably the heart of the book. Chapter 5, Sensational Content, causes trouble for the inferentialist view by attempting to isolate a base level content in sensation that does not depend on higher concepts, based on discussions by Christopher Peacocke and Allan Millar. The next chapter is on Kant and rather than settling for a treatment of Kant as a clear inferentialist, Davies favours an interpretation of his views in which he had doubts as to whether there is a necessary connection between experience and the faculties. The very possibility of experiential content that doesn’t involve high level concepts is congenial to the continuum account. Davies suggests that on the traditional picture of Kant it is unclear how sensation can link us with the world. Maybe that complaint assumes a kind of realism that is alien to Kant’s metaphysics. Chapter 7 defends the view that animals not only are conscious but have low-level, sensory concepts and thus have contentful sensations, and indeed have low-level, sensory, beliefs. This is in line with the continuum account.

Chapter 8 concerns Fodor’s modular account of perception (within an information-theoretic approach) in which input mechanisms, which have a reflex character rather than an inferential character, are distinguished from central processing which precisely does have the inferential character. Fodor argues that the persistence of, for example the Muller-Lyer illusion even when we are fully aware that the lines are in fact of the same length shows that there is a level of perceptual process that is not affected by inference from background beliefs. This provides Davies with support for his arguments against inferentialism and is integrated into the continuum account. Fodor’s view is linked with Davies’ favoured, non-standard, interpretation of Kant. In this chapter and the previous one on Kant, mention is made of Konrad Lorenz’s views on the evolutionary origin of perceptual structure. Without endorsing Lorenz’s views, Davies suggests that the felt qualities of sensations, by and large, confer evolutionary advantage by processing information in ‘conceptual shorthand’. Epiphenomenalism as a general view of sensations is rejected.

Part 3, Experience and Science, consists of a single chapter on Feyerabend. Davies rejects Feyerabend’s theory laden account of observation language but without endorsing the positivist position that the meaning of observation terms are given in experience. Consistent with material that came earlier, Davies favours the line that observational experience need not be inferential even if observational language is, and suggests Feyerabend conflates the issue of the nature of experience with the issue of the nature of observation terms. Davies defends the intuitive view that theory-less animals and children have experiences, with a low-level content, and places that in the context of his continuum view.

Part 4, Experience and the Mind, (chapters 10 – 13) starts, in chapter 10, with a discussion of Sellars’ attack on the ‘myth of the given’. Davies endorses Sellars’ conclusion that a non-inferentially known sense-datum is a myth, but doesn’t think that all givens have the characteristics of sense data. In particular, a low level awareness not of propositional form does not. Sellars’ argument involves pointing out that sense data need to be regarded as particulars if they are to be sensed without any acquired ability, but they need to be regarded as facts if they are to be the foundation of knowledge, and knowing a fact involves concepts, which are acquired. Davies’ position is that, given the ‘graduationist’ nature of his position, there doesn’t have to be full conceptual content with an inferential character for there to be some form of primitive content. Empirical evidence of ‘pictorial buffers’ is cited (p. 251) in support of Davies’ position in contrast to Sellars’ insistence on the propositional character of experiential content.

Chapter 11 concerns Paul Churchland’s eliminativism with respect to folk psychological terms, including terms which describe our sensory experiences. This is of concern to Davies because, for Churchland, such expressions should be eliminated because they do not refer, being part of a bad old theory. This is bad news if we want to defend an account of the content of experiences, as Davies does. Churchland is a very explicit inferentialist with perceptual judgments highly dependent on the nature of the overall theory in a way influenced by Kuhn, Davies explains. Davies argues that Churchland has not shown that just because theoretical terms are theory dependent that the content of observational experiences is theory dependent. This fifty page chapter deals with Churchland’s views in considerable detail.

Chapter 12 concerns Thomas Nagel’s views on the subjective and objective. Davies is interested in Nagel’s theme of the irreducibility of the subjective to the objective, and favours the version in Nagel’s writing where this suggests a dual aspect theory rather than pointing to the need for expanded conceptions of the mental and physical. Davies links Nagel’s dual aspect view with his own view “where certain contents of one’s perspectival experiences contain aspects which escape being captured at higher informational levels.” (p. 321) Chapter 13, the short final chapter, attempts to wrap up the main theme of the book in relation to Sellars’ two ways in which man relates to the world: employing either the manifest image or the scientific image. Davies’ view is that the typical materialist attempt to reconcile the two images, involving a reduction of the manifest image ultimately to physical terms, makes the same kind of mistake made by the inferentialist, of attempting to fully capture experience by descriptions.

The book retains a lot of the character of its origins as a (surely excellent) doctoral thesis (from Flinders University). It is very rich in its discussion of the literature and is careful, though also bold, in exegesis. At the same time, it develops a consistent overarching view. The breadth of material considered, which is a real strength, could also be viewed as a weakness in that the book’s very scope, length and attention to detail make considerable demands on the reader. There must be a good way to develop the main theme in a briefer, punchier form. But the work as it stands is a rich resource and, after an initial introduction to the main themes, could be read in selected parts. I learned an enormous amount from it.

Dr David Lumsden, University of Waikato
Australasian Journal of Philosophy Vol 76, No. 3, September (1998): pp. 508-510

Back to top

Review 2: Vladimir B. Popescu

W. Martin Davies, in his Experience and Content: A Continuum Theory presents a detailed and sustained argument for a thesis as to how the contents of experience, as a natural phenomenon, are fixed. More precisely, he advances a taxonomy of the contents of experience, and a view of the relations between them, with respect to the various roles that such contents may play in a cognitive economy. Davies’ project is thus distinct from that of explaining the content of experience in the sense of saying, in ways that don’t presuppose the notion of information, what makes experience contentful or informational. Davies’ concerns are, in fact, prior to the latter project. This much is evident from the fact that he follows connections between conceptions of experiential content and certain issues in epistemology, ontology, philosophy of science, and philosophy of mind. Consider some of the major claims Davies makes in this regard. On the view of experiential content he advocates, there are observationally fixed beliefs that have a constraining role in the acquisition of theoretical knowledge. But this does not entail a commitment to the discredited doctrine of foundationalism with which positivists, among others, were associated. The account also contains a robust distinction between observational contents of experience and those that are penetrated by theory. Furthermore, it supports the view that some contentful aspects of experience are subjectively felt properties of complex, evolved, physical systems. Whilst these properties are irreducible to and non-eliminable by the physical they may, perhaps, be causally efficacious.

Davies distinguishes his own continuum theory of content from what he considers to be the traditional alternative approaches, namely what he calls the inferentialist proposal and the observational account. The inferentialist claims that the content of any experience is exhausted by what judgements you can bring to bear on the contentless material provided by your senses. These judgements originate in all of the background linguistic-theoretical framework at your disposal. Such a rich background is considered to be necessary for the fixation of any experiential content. On the observational account, all of the contents of any experience are fixed non-inferentially, in a fast and reliable manner, by the observational situation with which you are presented. Davies’ continuum theory strikes a balance between these two accounts. Experience, Davies contends, can be either purely observational or it can contain a complex amalgam of contents, some of which are fixed non-inferentially by the observational situation, and some of which are fixed via inference from a linguistic-theoretical background. In his view, though all experience has observational contents, not all experience has inferential contents. Davies also holds that inference from background linguistic-propositional and theoretical information influences experiential content by degrees and not in an “all or nothing” fashion as the inferentialist claims.

The inferentialist proposal, Davies notes, has been historically the more influential doctrine. He claims, in support of this, that Immanuel Kant, Paul Feyerabend, Wilfred Sellars, and Paul Churchland are all inferentialists and provides detailed analyses of, and responses to, all of these thinkers’ views. To these I can not hope to do full justice here, but I will give an account of some of the crucial issues involved.

This dominance of the inferentialist thesis is rather strange in Davies’ view, given that, on the one hand, the observational account has certain strengths which are lost on the inferentialist account, and, on the other hand, the inferentialist account has a number of very counter-intuitive consequences, in spite of its strengths. It is clear for instance that the observational account offers the prospect of lining up issues of experiential belief fixation with evolutionary considerations which suggest that there should be fairly basic and fast mechanisms for the fixation of at least some beliefs. Such non-theory laden beliefs could provide a basis for distinguishing where rival theories might differ. By contrast, the inferentialist account, though it clearly captures certain contents of experience that can’t be captured by the observational account, can fairly naturally be seen to entail the notorious doctrine of the incommensurability of scientific theories, relativism, and a Kuhnian view of scientific progress as essentially a non-rational, social phenomenon. It also supports eliminativism in the philosophy of mind: Given the close relation between linguistic-theoretical practice and experience that inferentialists stress, someone like Churchland can go on to say that sensation itself is a contentless product of some background cognitive function, acquired through exposure to language and theory, and hence disposable by some other theory. For Churchland there is nothing left to constrain theory, at least in this case. Finally inferentialists would have to deny that developing infants and lower animals have experiences.

The explanation Davies offers for the dominance of inferentialism is that contemporary philosophers in particular seem to have assumed, mistakenly, that observationalism is necessarily linked to the discredited doctrines of empiricism and logical-positivism. The problem with these doctrines in the eyes of contemporary philosophers is, to put it rather coarsely, that whilst they can’t avoid being committed to observation statements, they still assume a robust distinction between observational and theoretical terms. But the latter is simply unavailable given that the meanings of terms in a natural language are not atomic, they are not fixed individually by some empirical situation, but are instead fixed holistically by a background of linguistic practice and theory.

It is clear, however, that these criticisms of the observational-theoretical distinction are concerned with the meanings of terms in a natural language. What has this to do with the contents of experience? Davies notes, with others, that the link to the observational account is achieved via a conflation of linguistic meaning with the content of experience which rests on a number of subtle confusions. For example, these critics of the observational account don’t seem to notice the important distinction between, on the one hand, experiential contents and, on the other hand, linguistic expressions or descriptions of experience. Another important distinction that goes unnoticed is that between experiential contents that are verbalisable and contents that are not.

To complete the case for inferentialism, as Davies sees it, notice that inferentialists tend to simply stipulate that the content of any experience must play a sophisticated epistemic role. Having made this stipulation they then point out that for an observational experience to have such a function it must be recognisable as having a meaning that can be communicated. Roughly then, the inferentialist case goes something like this: the epistemological role of experience requires that the content of any experience be linguistic-propositional; the assumed necessary link between language and the content of experience secures the thesis that experience has only linguistic-propositional contents anyway; given the collapse of the observation-theory distinction, as witnessed in the case of sense data theory and positivism, it follows that inferentialism and not observationalism is the correct account.

Now, Davies’ case that experience can be a complex amalgam of radically different contents where linguistic-theoretical background fixes experience by degrees, starts from a criticism of the inferentialist proposal but from a perspective which accepts many of the points that the inferentialist makes. Crucially I think, he accepts the rather strong thesis that any language-like propositional content of experience can be expressed verbally in the language of belief, and that these contents are fixed inferentially from a background of linguistic-theoretical information.

We find this strong assumption at work in a number of places in Davies’ argument for the continuum theory. His main point against the inferentialist is that developing infants and lower animals undergo experience in the absence of inference from a background of theory and linguistic practice. But there are two conclusions that one may try to draw here: a stronger and a weaker one. The strong conclusion is that such counter-examples signal the presence of radically different, non-propositional, contents. The weaker conclusion is that these contents are simply not determined by inference from background theory and are not verbalisable. And whilst the strong conclusion follows only if the strong thesis I mention is assumed, the weaker conclusion seems to be consistent with a view on which developing infants and lower animals can have experiences whose contents are the propositional contents of an innate language of thought, in Jerry Fodor’s sense, contents that are atomic and not holistic.

Davies also claims that we need to recognise both low-level informational but non-representational look ‘beliefs’ and high-level representational/propositional beliefs as aspects of an experience, where the red look of a tomato you see, for example, is distinct from your believing that the tomato is red but is nonetheless a part of your experience of the tomato. The contents of looks are non-representational, in Davies’ view, in virtue of the fact that they are not propositional. They do not constitute information that can be true or false. Such look beliefs are fixed by low-level, non-representational, but contentful, sensations. Now, Davies has fairly plausible things to say about the latter, independently of how we construe the contents of sensations. He claims that they are aspects of experience which are impenetrable by what we may call the high, personal level of belief, knowledge, and theory. This allows him to make a case for identifying them with informational states in Fodorian-like perceptual modules (species- and domain-specific information-processing mechanisms dealing with incoming stimuli from the senses). This helps him give support to the view that sensations may represent good candidates for constituting a basic ‘given’ in experience. Accepting this certainly rules out inferentialism. The anti-representationalist conclusion about sensations, however, follows only on the assumption that representation occurs only at the personal level. An innate language of thought hypothesis is consistent with sub-personal, impenetrable, representation, and we may identify low-level sensation with sub-personal representation in a language of thought. It is clear again that strong assumptions about the relationship between propositional representation and the level of verbalisable contents, knowledge, and inference from theory are at work.

It is only after establishing that low-level, non-representational, sensation is an experiential primitive, whereas high-level inferential content is not, that Davies advances evolutionary considerations in support of the continuum theory. For instance he can then, on this basis, argue that it is precisely the mechanisms responsible for the generation of sensation in response to physical stimuli on the senses, that are subject to evolutionary pressures in virtue of which they become epistemologically significant. Sensation, on Davies’ view, originally may have a causal but certainly not an epistemic role: Contrary to the inferentialist, some experiential contents at least need not play an epistemic role but they can and do sometimes. It is primarily this point about the evolutionarily determined constraining role of non-representational sensation in relation to high-level inferential contents that secures his claim that experience always has sensational aspects. He points out that we can’t expect any radical discontinuities in the experiences of organisms that are closely situated on the phylogenetic tree. The way to make sense of the increases in sophistication in experience as we move up the tree is in terms of viewing experience as an amalgam of sensorial and inferential contents, where the latter occur in increasing degrees of sophistication.

To conclude, it should be clear from what I say here that Davies’ book is impressively wide-ranging, and should be of interest to anyone concerned with how issues in epistemology, philosophy of science, and philosophy of mind may be connected.

Vladimir B. Popescu
University of Adelaide, Metascience June (1998): pp. 144-148.

Back to top

Review 3: Professor Andrew Latus

Davies presents a theory of experiential content which he calls the ‘continuum theory’, a theory that occupies the middle ground between two opposed accounts of content. On one side we have the inferentialist account, according to which inferences from ‘sophisticated high-level knowledge’ is necessary (or on a strong view, necessary and sufficient) for the having of experiences (19). On the other side, is the observational account,according to which ‘only the observational situation is relevant to how experiences originate’ (20). However, although Davies introduces his account by contrasting it with both these accounts, his real target is the inferentialist account. Davies convincingly argues that something like this account lies behind most contemporary accounts of content. He finds it, for instance, in the work of Churchland, Armstrong and Harman. It is not clear, on the other hand, that anyone endorses the observational account. The observational account is important only as a device for staking out the theoretical territory.

Davies’ own account tells us that, while sophisticated high-level features very often do play a role in the having of an experience, they need not. According to Davies, content comes in various sorts ranging from high-level to low-level and inference becomes less important as the level of content concerned diminishes. At the low-level, inference is unnecessary for content. Most experiences combine various sorts of content from high-level to low and so require inferences to be made for the experience in question to occur. Nonetheless, it is possible to have experiences which consist of nothing but low-level content and so do not require any inferences from high-level factors to have been made.

All this talk of high-level and low-level factors probably seems quite mysterious. Davies is slow to make the distinction clear. At first he uses the terms without explanation. Then, a little way into the book, he lists some of the factors he considers high-level: ‘concepts, theories, background knowledge’ (27). This seems helpful and these are surely factors that figure in inferences, but he soon muddies the waters again by revealing that there are also low-level versions of some of these factors (54-60). Some concepts, for instance, are not ‘descriptive’ (i.e., high-level), but ‘sensational’ (i.e., low-level). What marks the difference between these sorts of concepts? The answer has to do with language. High-level concepts are expressible in language, low-level concepts are not. Instead, low-level concepts are ‘best characterised in sensational terms’ (56). What exactly sensational concepts are then is not clear. If they are expressible in language it is difficult to know what they are and what Davies has to say about this is not particularly helpful. We are best off to forget about the different sorts of concepts and instead focus on the connection between high-level factors and language, for this connection is at the heart of Davies’ position. The core of his position is revealed in the claim that ‘experiences are not exclusively language-like’ (69). Davies claims that, to the extent that experience is language-like (i.e., expressible in language), high-level factors such as descriptive concepts are necessary for experience, but since experience is not entirely language-like, high-level ‘linguistic’ factors are not absolutely necessary for experience.

Since Davies presents the inferentialist account as the orthodox theory of content, an important task he faces is that of giving us reason to think the inferentialist account is wrong. His attempts to do so are the most frustrating part of the book. Davies’ attacks on inferentialism often boil down to the claim that it is obvious his opponents are wrong, for instance, that it is clear ‘high-level features … are not always present in experiences of unsophisticated creatures like animals and infants’ (59). If this is truely obvious then clearly the inferentialist account fails, for if so, unsophisticated animals sometimes have experiences without inference from high-level factors. But it is unreasonable to think that such appeals to obviousness have any hope of establishing the inferentialist proposal is false. One not clearly, although probably distasteful, response to response to Davies is the claim that Descartes was right, unsophisticated animals don’t have experiences at all. Furthermore, even if this response is unappealing, we may respond that creatures have experiential capacity in proportion to to their inferential capacity. If this is right, we are not forced into the view that animals have no experiences at all, they simply have more limited experiences. Davies is sometimes guilty of ignoring this option, attributing to the inferentialist the view that, since it is not terribly plausible that animals are as good at making inferences as we are, they must have no experiences at all. For instance, he suggests at one point that, on the inferentialist view, ‘since unsophisticated animals are said to have few inferential mechanisms at their disposal, then they must have no experiences’ (169, emphasis mine). Not so, they must just have more limited experiences.

I shouldn’t, however, give the impression that Davies engages in sloppy arguments throughout his book. The book is filled with much close and careful argument. Davies only becomes careless when it comes to this fundamental point about whether experience can consist entirely of low-level non-inferential content. The problem is that, for any case in which an anti-inferentialist claims there is clearly experience without inference, it is always open to the inferentialist to respond that the case is actually one in which some sort of inference was made or in which no experience occurs. Decisive arguments against either of these responses have yet to be discovered. Davies’ case for the continuum account is much stronger when he makes a positive case for it. He stresses, for instance, how well it meshes with the modularity of mind thesis and how well it allows us to tell an evolutionary story about the emergence of consciousness. These are both good reasons for giving his view serious consideration.

Although Davies fails in one of his central aims, there is much to recommend about his book. It is extraodinarily ambitious. He uses his account to consider a far ranging array of topics including Kant on cognition, Kuhn on theory change, the philosophy of colour, property dualism, Fodor on the modularity of mind, Churchland on eliminative materialism, Sellars on the myth of the given and the scientific image, and Nagel on what it is like to be a bat. Much of this discussion is good and some of it, particularly the chapters on Fodor and Churchland, is excellent. He is sometimes not as clear as he might be, but much of the book rewards careful reading.

Professor Andrew Latus, University College of Cape Breton
Philosophy in Review Vol 18, No. 2, April (1998): pp. 92-94

Back to top