: Content, Functional Mapping, and the Myth of the Given
Just hat are e tal!ing a"out hen e tal! a"out the content of perception? There are a num"er of different #uestions hich are often as!ed a"out the precise nature of this content, such as to hat e$tent it is representational, and to hat e$tent this representation is conceptual in character% &oever, these #uestions usually gloss over hat it is to tal! a"out perceptual content qua content, in their haste to tal! a"out it qua representation or conception% 'n this "asis, it is all too tempting to account for not only the specific character of the content, "ut also its mode of individuation, in phenomenological terms% This ma!es perceptual content into hatever is contained ithin an introspective domain to hich e have some sort of special access% (ilfrid )ellars* attac! on the myth of the given has provided us ith good reason to dou"t the epistemic authority such special access to our inner states purportedly provides, and thus to dou"t the efficacy of any account of perceptual content that gives pride of place to introspection% This is his criti#ue of hat Jim '*)hea has called the epistemic given% The aim of the present paper is to shed some light on his criti#ue of the other side of the myth, or hat '*)hea calls the categorial given, "y or!ing through some of his other ideas% This involves demonstrating constraints governing accounts of perceptual content on the "asis of the explanatory demands placed upon them as accounts of perception, and the explanatory resources availa"le to them as accounts of content% +% ,erception and -$planation To "egin ith the e$planatory demands, it seems to me that there are to distinct e$planatory enterprises that any account of perceptual content must contri"ute to% 'n the one hand, there is the epistemological enterprise of e$plaining the general role that perception can play in empirical .ustification independently of variations in the causal structure of perceiving agents% This means telling a story a"out ho the sensory inputs fed into a causal system can gain the normative significance of arranting moves ithin the space of reasons, in a ay that could apply not only to different human "eings, "ut to stranger creatures such as aliens or artificial intelligences hose sensory capacities and overall causal economy diverge radically from our on% 'n the other hand, there is the psychological enterprise of e$plaining the specific role that perception plays in the causal economy of particular perceiving agents% This means telling a story a"out ho the sensory inputs fed into a causal system contri"ute to the production of behavioural outputs, in such a ay as to give us predictive purchase upon the "ehaviour of creatures ith particular types of causal structure% Insofar as introspection is a further !ind of perception, it is something that needs to "e + accounted for in "oth stories, and this gives us reason for caution in using it to tell these stories% /lthough these enterprises are distinct, they are also importantly intertined% /ny epistemology that cannot account for the ay differences in the causal structure of our sensory capacities can affect their role in empirical .ustification ill have failed to get a grip on the causal dimension of perception, or its connection to sensation, and any psychology that cannot account for the ay perception can supply us ith reasons that cause us to act one ay rather than another ill have failed to get a grip on the normative dimension of perception, or its connection to rational agency% (e only have an account of perception, "e it epistemological or psychological, hen its causal and normative dimensions are properly connected% /ny account of perceptual content as something that plays a role in "oth epistemological and psychological e$planations must couch it in terms amena"le to "oth of these dimensions% The first insight that e can ta!e from )ellars here is that the proper interface "eteen these to dimensions is the use of functional explanation in empirical psychology% To use a phrase he is fond of 0 e must understand perceptual content in terms of the ay it fits into the wiring diagram of the perceiver% Causal e$planation in general or!s "y applying explanatory schema to systems that facilitate the development of predictions a"out the ay that they ould "ehave under various possi"le conditions% These schema provide us ith more or less general ays of organising counterfactual reasoning a"out these possi"ilities, there"y ena"ling us to dra specific conclusions a"out ho they ould "ehave in any given set of circumstances% / functional schema ena"les us to develop predictions "y treating a system in analogy ith practical reasoning% For instance, "y alloing us to treat its parts as means in relation to the whole as an end% This lets us descri"e the causal role of the systems* components in terms of success and failure, and there"y to organise our counterfactual reasoning a"out the causal relations "eteen them in terms of the ay failure cascades throughout the system% The e$planatory poer of a functional schema thus lies precisely in its introduction of the possi"ility of malfunction% Crucially, this is an essentially normative notion, hich can "e leveraged in connecting the epistemological and psychological accounts of perception% 1% Content and -$planation Moving on to the e$planatory resources, it stri!es me that the purpose of the notion of content is to tal! a"out ho to seemingly different states of distinct systems can nevertheless "e the same in another sense% There*s o"viously a trivial sense in hich this can "e the case ithout arranting the 1 ascription of content at all, such as the sense in hich to houses may "e in a state of disrepair, even hile the precise nature and e$tent of their disrepair may differ% 2ut to say that they "oth share a content here ould "e to say nothing more than that they share a property, and thus entirely redundant% This is not to say that states hose similarity consists in common properties cannot share content on that very "asis% To V&) tapes that share precisely the same magnetic properties ill share precisely the same informational content% The #uestion is ho e can say that an information storage medium that or!s on different causal principles, such as a 2etama$ tape or a 3V3 could share the same content as the V&) tapes% I*m going to use this comparison ith information storage media as the guiding analogy through hich to thin! a"out the e$planatory role of perceptual content% I thin! e should aim to thin! a"out hat it ould "e for states of distinct perceiving su".ects qua causal systems to have the same content, in much the ay that states of distinct information storage devices can have the same content, "efore e address hat it ould "e for this content to "e perceptual, and in hat sense this ma!es it representational and4or conceptual% In order to e$tend the individuation of the content "elonging to states of causal systems "eyond mere similarity of properties, e must consider relations of isomorphism "eteen them% This means "rea!ing don states into the features of hich they are composed and the relations "eteen them, and then developing a ay of mapping these to the features and relations that compose the corresponding states% This mapping schema 5or morphism6 allos us to count some feature of one state as e#uivalent to a feature of another despite differences in their properties, insofar as it occupies the same role ithin the system of relations that constitute it% (e can then determine if to states share the same content on the "asis of some sufficient degree of correspondence "eteen their components% This ould let us see a V&) and a 2etama$ tape as possessing the same content insofar as there is some ay of mapping the magnetic properties of one to the other, despite the differences "eteen these properties% &oever, the pro"lem ith such pure isomorphism is that it can "e ar"itrarily e$tended in ays that undermine any possi"le e$planatory role it could have% (e can potentially construct ar"itrary mappings that pair the magnetic traces on the V&) tape ith price patterns in the stoc! mar!et, or pair the digital encoding of the 3V3 ith a se#uence ta!en from the "inary e$pansion of pi% Insofar as it completely severs the individuation of content from any concern ith the causal capacities of systems and their states, pure isomorphism precludes identity of content from playing any role in causal e$planation% (e can avoid these pro"lems of pure isomorphism "y using the functional roles of the states and their components to constrain the mapping schema% In order for this to or! the relevant states must 7 "e variable features of the iring diagram of the system, and their variations must "e functionally correlated ith variable outputs of some sort% For e$ample, the V&) tape contains a length of material hose electromagnetic properties vary in delimited ays, hich ill produce suita"le variations in the patterns of light emitted "y a TV set to hich it is appropriately connected% 'n this "asis, it is possi"le to produce a functional mapping from the components of the one set of varia"le states to another in terms of common outputs to hich they are functionally correlated% For e$ample, e can produce a functional mapping from V&) to 2etama$ that maps their distinct variations in electromagnetic properties onto one another in terms of similarities in the patterns of light they engender hen suita"ly connected to a TV set% In essence, e treat the states possessing content as isomorphic ith one another insofar as their components can "e mapped onto the same set of functional roles, hich are themselves isomorphic ith the output they are supposed to produce% There are some issues here regarding the fineness of grain of this isomorphism, "ut e ill leave them "e, in favour of to final points a"out functional mapping% The first point is that e need not characterise content individuated this ay as representational% It is all too tempting to say that the content represents the output that it is functionally correlated ith% For instance, it can "e tempting to suggest that to V&) tapes represent the same movie, or that to records represent the same piece of music 5something )ellars is guilty of at one point6% &ere it is important to remem"er that 8movies* or 8pieces of music* are .ust as abstract as contents, and their correct shoing or performance is su".ect to their on set of criteria% 'ne might nevertheless thin! that content represents raw output, such as the patterns of light or sound ith hich the information storage media are correlated% &oever, e should resist the temptation to identify representation ith mere functional correlation, as it arises principally "ecause the analogical character of functional e$planation invites us to treat the relevant states as if they ere instructions for performing certain actions% The second point is that this gives us a ay of tal!ing a"out form as ell as content% The ays in hich the components of the relevant states can vary may "e classified in terms of more general functional roles they play, and these classifications provide more or less a"stract forms that corresponds to the content of their specific variations% ,ut another ay, form consists in the functional invariants that delimit the variations in hich content consists% It is the structure of those elements of the iring diagram in hich the varia"ility of the content "earing states are encoded% 9 7% ,erception, :epresentation and Conception (e can no turn to considering hich states of a causal system deserve to have something called perceptual content ascri"ed to them, and to hat e$tent they are representational and4or conceptual% &ere I am going to follo )ellars, ho ta!es perception to "e the transition from a causal input to a conceptually articulated output, or from sensation to conception% I*m also going to endorse his account of the nature of this conceptually articulated output, hich is thought of as a sentence or as a state hose content is e#uivalent to a sentence% Conceptual content is principally understood as the functional role that a sentence or its component e$pressions play in the game of giving and as!ing for reasons, hich is composed from three types of move that are made ith sentences: language-entry moves 5perception6, intra-language moves 5inference6, and language-exit moves 5action6% To sentences have the same content .ust insofar as the same moves can "e made ith them, and to component e$pressions have the same content .ust insofar as their contri"ution to the roles of the sentences they compose is the same% Conceptual content can "e derivatively possessed "y states of systems insofar as they are appropriately functionally connected to the possi"ility of linguistic output, i%e%, of actually ma!ing moves ithin the game% This means that the sense in hich the transition from a causal input to a conceptually contentful state counts as perceptual is derivative upon the sense in hich a language;entry move is perceptual% Finally, I*m going to say that conceptual content is representational insofar as I agree ith )ellars that it signifies things in the orld, "ut I*m not going to go any further into the tortured #uestion of ho representation can "e reconstructed out of inference here% 'n this "asis, I thin! that anything orth the name perceptual content ill "e have to "e possessed "y states of the mechanisms involved in the hole process of moving from sensation to conception% 2orroing a term from 2random, e could say that perceptual content must "e possessed "y some state of the perceptual mechanism underlying a rational agent*s reliable differential responsive dispositions 5:3:3s6% &oever, there are potentially many candidates for this, insofar as there can "e numerous su"system states ith varia"le functional outputs involved in the processing of sensory information into conceptual content% :estricting ourselves to our on visual systems for the moment, there is everything from the pattern of activation on the "ac! of the retina 5<uine*s infamous stimulus meaning6, through states of information processing systems in the visual corte$, to states of a glo"al system that integrates information from various sources and ma!es it availa"le to other cognitive processes 5such as Thomas Met=inger*s Phenomenal World Model6% Moreover, "ecause content "earing states compose into further states, it is e#ually possi"le to tal! a"out > com"inations of any or all of these% There are multiple layers of information processing "eteen sensation and conception any and all of hich can "e the su".ect of functional mappings insofar as they output to other layers or to the ultimate conceptually articulated product% The #uestion is no hat it ould "e to say that any of these states possessed representational yet non;conceptual content% This is hat )ellars calls picturing as opposed to signifying% The o"vious thing to do here is to reach for the notion of isomorphism once more, and to say that content hich is already individuated "y functional mapping is representational .ust insofar as there is also an isomorphism "eteen it and some state in the orld, hich it is there"y ta!en to represent% For e$ample, the content of the V&) and 2etama$ tapes may "e the same insofar as they record the data from the same security camera, and they may then "e ta!en to represent the same events insofar as there is an isomorphism "eteen this content and the relevant events% &oever, this suffers from the same pro"lems ith ar"itrary mappings e discussed earlier, insofar as e can con.ure up isomorphisms ith other potentially stranger events or states% (e might then suggest that this isomorphism is constrained "y the causal origin of the relevant states% &oever, this ould leave us saying that if the tapes ere arped in precisely the same ay "y the same magnetic field, that they there"y represent that magnetic field% )ellars* solution to this pro"lem is that the isomorphism must "e more deeply tied into the functional role of the content "earing state% In 82eing and 2eing ?non*, he illustrates this using the e$ample of a ro"ot that stores information on a similar magnetic tape, hich he ta!es to picture its environment in virtue of an isomorphism "eteen the state of the tape and the state of the environment% &oever, he also claims that: This picturing cannot "e a"stracted from the mechanical and electronic processes in hich the tape is caught up% The patterns on the tape do not picture the ro"ot*s environment merely "y virtue of "eing patterns on the tape% In (ittgenstein*s phrase, the 8method of pro.ection* of the map involves the manner in hich the patterns on the tape are added to, modified, and responded to "y the other components of the ro"ot% It is a map only "y virtue of the physical habitus of the ro"ot, i%e%, "y virtue of mechanical and electronic propensities hich are rooted, ultimately, in its iring diagram% 5@9A6 To summarise, the constraints upon an isomorphism that allos us to produce a representational mapping "eteen the states of to systems does not merely concern similarity of functional output, "ut a more comple$ relation of projection in hich the state mediates "eteen functionally B specified inputs and outputs% For e$ample, e can understand a varia"le state of the iring diagram of a "ee as representing the path to a source of nectar insofar as it varies appropriately hen the "ee discovers the source of nectar and produces the "ehaviour of returning to that source of nectar ith other "ees that it has danced for% This is something that :uth Milli!an has e$panded upon in much more detail% )o, for a state of our perceptual mechanisms to "e non;conceptual and yet representational in this sense ould "e for it to play a functional role in a process of systematically guiding behaviour in relation to sensory input that is to some e$tent independent of any role played "y conceptual content in guiding action, hich is derivative upon the role of sentences in language;e$it moves in much the ay that perception is derivative upon the role of sentences in language;entry moves% The tension here, hich I do not aim to resolve, is the e$tent to hich these states can play an active role in the move from sensation to conception ithout their role in guiding "ehaviour "eing similarly "ound up in conception% 9% Cniversal and ,arochial Content &aving distinguished "eteen the non;representational, non;conceptual, and conceptual contents that states of our perceptual mechanisms can "ear in terms of the functional roles of their components ithin these mechanisms, I no ant to ma!e a further distinction "eteen types of content on this "asis, and see ho this distinction can shed some light on the myth of the categorial given% The distinction I ant to dra is "eteen hat I*ll call parochial and universal forms of content% The former covers all forms of content hose individuation is dependent upon functional mappings that are specified in terms of particular causal mechanisms, hereas the latter covers all forms of content hose individuation is entirely independent of any particular causal mechanism% (e*ve already made this sort of distinction in considering the difference "eteen contents individuated in terms of functional mappings and contents individuated in terms of pure isomorphisms% The characterisation of the latter is entirely mechanism independent, "ut ultimately cannot play a useful role in any causal e$planation, psychological or otherise, hereas the characterisation of the former, along ith the representational mappings e have discussed depends upon the causal structure of the mechanisms in terms of hich the output is specified% &oever, the conceptual content hich is the result of perception is universal in the sense .ust defined% This is "ecause the language games from hich the conceptual content of states is derived D are social practices that are intrinsically extensible% Though any given spea!er*s a"ility to perform language;entry moves is tied to the structure of their on perceptual mechanisms, the concepts that they there"y apply are not forever tied don to those mechanisms% It is entirely possi"le for spea!ers ith entirely distinct causal economies to ac#uire the concept as long as their linguistic usage can "e appropriately mapped from ithin the game of giving and as!ing for reasons itself, insofar as the causal structure of any given perceptual mechanism is something amena"le to analysis in conceptual terms% This is the same process through hich e incorporate ne and more accurate measurement devices into our practices, there"y "ootstrapping our a"ility to apply empirical concepts on the "ac! of our parochial perceptual mechanisms% Eevertheless, this e$tensi"ility and the universality it implies does not detract from the fact that our concepts are achievements of distinct sort% Concepts display relatively fi$ed forms of functional invariance in much the ay that forms of functionally individuated content do% This is hat )ellars* means "y the categorial% Form in general is the classification of functional roles% Categorial form is the classification of specifically conceptual roles% Categories are thus general classifications of concepts that constitute the general form of signification as opposed to the general classifications of causal states involve in the various parochial forms of picturing% /ll this gives us a ay of thin!ing a"out the illusion that engenders the myth of the categorial given% I*ll #uote )ellars* on this form of the myth: To re.ect the Myth of the Given is to re.ect the idea that the categorial structure of the orld 0 if it has a categorial structure 0 imposes itself on the mind as a seal imposes an image on melted a$% 5@9>6 I thin! e can reformulate the myth as follos: it consists in the idea that there is some form of universal perceptual content that is distinct from empirical conceptual content as )ellars descri"es it% This is .ust the claim that there is some specific sense in hich a"solutely any to sentient creatures could "e said to have the same experience ithout having the same conceptual grasp of this e$perience% This amounts to the idea that I have a TV in my head that has no particular causal; functional structure% It is the postulation of a !ind of content that cannot in principle "e individuated, or rather, hich is self;individuating% This derives from the temptation to individuate perceptual content on the "asis of what e ta!e it to represent 5e%g%, the pin! ice cu"e e*re "oth loo!ing at6, rather than how it represents it 5e%g%, the functional role the relevant states play in a F ider "ehavioural economy that incorporates the pin! ice cu"e6% This ma!es the posited content universal, "ut only insofar as it "ecomes parasitic upon the conceptual content of sentences e use to descri"e hat it represents% Cnder the poer of this myth, the appearance of some common form ithin introspective apprehension of this purportedly universal content there"y engenders the illusion that the orld itself is impressing its categorial structure upon us directly% G