What do you think?
Language, thought, and other biological categories: New foundations for realism
Preface by Daniel C. Dennett
Beginning with a general theory of function applied to body organs, behaviors, customs, and both inner and outer representations, Ruth Millikan argues that the intentionality of language can be described without reference to speaker intentions and that an understanding of the intentionality of thought can and should be divorced from the problem of understanding consciousness. The results support a realist theory of truth and of universals, and open the way for a nonfoundationalist and nonholistic approach to epistemology.
Ruth Millikan is Associate Professor of Philosophy at the University of Connecticut at Storrs. A Bradford Book.
Beginning with a general theory of function applied to body organs, behaviors, customs, and both inner and outer representations, Ruth Millikan argues that the intentionality of language can be described without reference to speaker intentions and that an understanding of the intentionality of thought can and should be divorced from the problem of understanding consciousness. The results support a realist theory of truth and of universals, and open the way for a nonfoundationalist and nonholistic approach to epistemology.
Ruth Millikan is Associate Professor of Philosophy at the University of Connecticut at Storrs. A Bradford Book.
355 pages, Hardcover
First published January 1, 1984
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 6 of 6 reviews
June 18, 2026
Ruth Millikan offers a common-sense argument about how we mean things with language and why those meanings bring about change in the world. The core of her argument is that language use is purposive and functional. When we use language, it is with intention. For those intentions to be fulfilled, language must map onto a world of shared experiences and do so in conventional ways so as to induce a cooperative relationship between communicator and interpreter. Our success in fulfilling intentions depends on whether our efforts in language are recognizable and sufficiently mappable to the circumstances of interpretation.
I like this argument because it does not imply that meaning is always constructed anew in each circumstance of use. It also subverts any belief that language simply means what we choose it to mean. I also like that the argument is extendable in ways that other analytic work on language is not. Although Millikan focuses her arguments and examples on isolated sentences, the principles can be adapted to language use in streams of discourse to which it must cohere without losing its correspondence to a reality outside of our minds.
Ultimately, Millikan is most concerned with grounding intentional language use in a framework of realism. Functional and conventional uses of language, she argues, can be likened to biological adaptations. In fact, the reason for those adaptations (i.e., survival) are the same. There are two key concepts in this argument.
Biological Categories are adaptations in an organism that result from lived engagement with reality. The development of protective coloring, means of locomotion, apparatuses for breathing, etc. are adaptations that ensure survival. Survival effectively selects the biological adaptations that continue through the generations. Words and the language constructions we put them into are similarly adaptive. We keep words that effectively map to things in the world and keep language constructions that result in our intentions being reliably fulfilled.
Teleosemantics is the idea that we use words to point to the world and use them to create relationships with the world and others in it. When we are successful in doing so (e.g., I speak and am understood) then that success helps select that expression or word for continued use. Over time, the meaning of gestures, signs, expressions, and even the words themselves come to mean what it is useful for them to mean, and they convey that information in forms that are most likely to be recognized and to map onto a random interlocutor’s “sense” of the world. This slate of conventional language units can be adapted within the range of typified use to bridge over local variance in our circumstances of interpretation.
I see this as a genre-based view of language, where our available language choices derive from a history of language use in which successes and failures if communicating intention have left us with conventional forms of expression to meet conventional and recognizable occasions for language use. By continuing to use language in conventional ways, we lend stability to our interpretations and relationships. As those circumstances changes, our language does as well, adapting to new purposes, but eventually re-settling into what Millikan calls “Normal” use.
For Millikan, the ability to think thoughts and to rationalize their logical possibilities does not mean that they correspond to reality. This is a “meaning rationalism” position that she accuses Quine, Wittgenstein, Hume, and Putnam of holding. She insists that meaningful and purposive language use must correspond to reality and do so in a way that preserves our “sense” of the world, the way that we grasp it and act on it, individually and as social beings. These positions are not incompatible with one another.
The book is a bit dense, perhaps necessarily so, given the work she is engaging with and responding to. However, the argument is quite clear and compelling.
I like this argument because it does not imply that meaning is always constructed anew in each circumstance of use. It also subverts any belief that language simply means what we choose it to mean. I also like that the argument is extendable in ways that other analytic work on language is not. Although Millikan focuses her arguments and examples on isolated sentences, the principles can be adapted to language use in streams of discourse to which it must cohere without losing its correspondence to a reality outside of our minds.
Ultimately, Millikan is most concerned with grounding intentional language use in a framework of realism. Functional and conventional uses of language, she argues, can be likened to biological adaptations. In fact, the reason for those adaptations (i.e., survival) are the same. There are two key concepts in this argument.
Biological Categories are adaptations in an organism that result from lived engagement with reality. The development of protective coloring, means of locomotion, apparatuses for breathing, etc. are adaptations that ensure survival. Survival effectively selects the biological adaptations that continue through the generations. Words and the language constructions we put them into are similarly adaptive. We keep words that effectively map to things in the world and keep language constructions that result in our intentions being reliably fulfilled.
Teleosemantics is the idea that we use words to point to the world and use them to create relationships with the world and others in it. When we are successful in doing so (e.g., I speak and am understood) then that success helps select that expression or word for continued use. Over time, the meaning of gestures, signs, expressions, and even the words themselves come to mean what it is useful for them to mean, and they convey that information in forms that are most likely to be recognized and to map onto a random interlocutor’s “sense” of the world. This slate of conventional language units can be adapted within the range of typified use to bridge over local variance in our circumstances of interpretation.
I see this as a genre-based view of language, where our available language choices derive from a history of language use in which successes and failures if communicating intention have left us with conventional forms of expression to meet conventional and recognizable occasions for language use. By continuing to use language in conventional ways, we lend stability to our interpretations and relationships. As those circumstances changes, our language does as well, adapting to new purposes, but eventually re-settling into what Millikan calls “Normal” use.
For Millikan, the ability to think thoughts and to rationalize their logical possibilities does not mean that they correspond to reality. This is a “meaning rationalism” position that she accuses Quine, Wittgenstein, Hume, and Putnam of holding. She insists that meaningful and purposive language use must correspond to reality and do so in a way that preserves our “sense” of the world, the way that we grasp it and act on it, individually and as social beings. These positions are not incompatible with one another.
The book is a bit dense, perhaps necessarily so, given the work she is engaging with and responding to. However, the argument is quite clear and compelling.
August 17, 2017
Whoa, I did it. I read this incredibly dense book reconstructing meaning, ontology, and epistemology upon materialist, realist terms. I had to start over at one point. But I made it and I'm glad I did. Not that the writing was bad, not at all. But the concepts were heavy-duty. Going to read more Millikan.
October 28, 2021
Millikan's book will be challenging to read for those unfamiliar with the topic (basically, meaning, mental representation), since she engages with various preceding philosophers, but it will also be challenging for those who ARE familiar with the topic, since her project involves re-defining some main concepts (e.g., 'intension,' 'intentionality,' 'representation,' 'concept,' 'belief,' 'substance,' 'property,' etc.) and introducing new ones, as well as a whole new direction of thinking.
Particularly for those interested in the topic (and who've read in this area), the book is worth the substantial amount of effort it demands--sort of. But it is just so much harder than it needs to be. She gets off track too much in debating previous philosophers--sometimes when she does this, it clarifies her own position, but I wouldn't say it usually works like this. For instance, her disagreements with Leibniz and Aristotle about properties is essentially to say they have a different account of what properties are-- but her account is complicated enough that it's just a useless distraction, since their views--Aristotle's in particular--are weird themselves. How much weirdness must you subject your reader to at once, Millikan?
Don't get too hung up on the details of the book. The most interesting insights will come through even if you don't know what Wittgenstein said wrong or whatever. And you can come back to them.
Particularly for those interested in the topic (and who've read in this area), the book is worth the substantial amount of effort it demands--sort of. But it is just so much harder than it needs to be. She gets off track too much in debating previous philosophers--sometimes when she does this, it clarifies her own position, but I wouldn't say it usually works like this. For instance, her disagreements with Leibniz and Aristotle about properties is essentially to say they have a different account of what properties are-- but her account is complicated enough that it's just a useless distraction, since their views--Aristotle's in particular--are weird themselves. How much weirdness must you subject your reader to at once, Millikan?
Don't get too hung up on the details of the book. The most interesting insights will come through even if you don't know what Wittgenstein said wrong or whatever. And you can come back to them.
June 17, 2026
Chapter 1: Direct Proper Functions: She attempts to explain how things can have proper functions unmysteriously. She first defines reproduction–B is a reproduction of A iff they have some determinate property p in common, and it is explained by natural laws such that if A changed in p B would as well. A reproduction can be a reproduction of many things. Reproductively established families are copied from the same thing. To be conventional, something needs to be from reproductively established family, contrary to being naturally produced. Higher-order reproductively established families include my heart and my parent’s heart because they were produced by something whose proper function includes making copies (genes). Additionally, it also includes things produced by trial by learning because there is a mechanism that is supposed to produce the same rewarded behavior as before. But she also wants to be able to include malformed things. Then she gives this important definition: “A function F is a direct proper function of x if x exists having a character C because by having C it can perform F.” She then previews how it can apply to language and defines Normal explanations, Normal conditions and focused proper functions.
Chapter 2: Adapted Devices and Adapted and Derived Proper Functions: She defines relational proper functions, adapted proper functions, adapted devices, derived proper functions, etc. all conditioned behavior, whether relational or nonrelational, is behavior that has derived proper functions. Then she tries to explain how learned behavior can have functions–they are derived proper functions because they are the result of systems that were evolved to do things and so who have direct proper functions. Additionally, mental states are kinds of adapted devices and then human actions derive functions from them.
Chapter 3: Indicatives, Imperatives, and Gricean Intentions: She argues that one aspect of the meaning of a word, syntactic form, or other language device is its focused stabilizing function. The full direct proper function of any sentence is derived from the stabilizing functions of its significant elements or aspects, each of which belongs to its own reproductively established family. The stabilizing function of the imperative mood is to produce compliance; the stabilizing function of the indicative mood is to convey information, i.e., to produce true belief. Against Grice, It is very likely that small children speak and understand long before they have any such concepts of belief and intention.
Chapter 4: Language Device Types; Dictionary Senses; Stabilizing Proper Function as the First Aspect of Meaning: Denotative and referential terms are most special not because of their function but their intentionality which has more to do with the way a device Normally performs
its proper functions than with its proper functions per se. They function properly when an interpreter identifies or recognizes what their referents are. What makes two sentence tokens
tokens that display the same syntactic form should likewise be a matter of history (what reproductive family it is from), rather than mere physical form or actual function or a speaker’s intentions. What determines what a sentence meaning is history–stablizing function.
Chapter 5: Intentionality as a Natural Phenomenon: This is mainly an introduction to the main ideas she will discuss in the next section.
Chapter 6: Intentional Icons: Fregean Sense, Reference, and Real Value Introduced: She first defines intentional icons as including as being a member of a reproductively established family, stands midway between a Producer and Interpreter such that the interpreter can carry out its proper functions. What they represent is determined by the last member of the series of things it is supposed to map onto and the most proximate Normal explanations, respectively. Next, she responds to the problem of how words and sentences can correspond to nothing real. In response, she answers that what a sentence means is dependent on its history, of what its mapping looks like in Normal conditions. Then she argues that the "sense" of a language element is its Normal mapping rule. Importantly, intension is not the same as sense. Then she discusses real value which an intentional icon only has when it maps onto what it is supposed to. Lastly, when a term or token has a referent, there exists an x such that the term or token is supposed to map onto x.
Chapter 7: Kinds of Signs: Argues that intentionality comes in degrees using examples such as pressure gauges, adrenalin, art, and improvised signs.
Chapter 8: Hubots, Rumans, and Others: Case Studies of Intensions, Senses, and '"Stimulus Meanings": In this chapter, she argues that neither the "stimulus meaning" of a term nor its explicit intension (“rational being” -> “human”) determines its sense. The sense of a term determines neither its "stimulus meaning" nor its explicit intension. Senses—even the senses of thoughts—can be ambiguous (if the reason that an intentional icon is there is disjunctive) and they (or the purported thoughts) can be vacuous (if there is no relevant history). Additionally, lacking an explicit intension that has sense, a vacuous term is no more intentional than a sneeze.
Chapter 9: Intensions: Intension is meaning at all only insofar as intensions govern terms that
have sense or insofar as intensions are explicit and themselves have Sense. Additionally, the act of identifying the referent of a term or of coming to be able to identify it is not the same as the vague affair "coming to know what the term means."
Chapter 10: Simple Indexicals: An indexical does not have a fixed referent or fixed sense on its own. Instead, it has a relational sense: a rule that tells us how to find its referent from its relation to the context of use. This means they have proper functions which require contextual adaptors. Their referents emerge from real-world relations between speaker, hearer, and environment.
Chapter 11: Descriptions: Indefinite descriptions are normally descriptive and do not genuinely refer. Definite descriptions are primarily devices for introducing referents, not uniqueness.
The sense of a definite description consists in the mapping rules that determine which individual or property it picks out.
Chapter 12: Identity statements are for telling speaker to merge two concepts. It is not a relation. Existential statements tell the hearer to engage reference, and nonexistential statements the opposite.
Chapter 13: Explains typing, displaced function, and context-sensitive preservation rules in terms of what they’re supposed to do.
Chapter 14: Negative sentences correspond to positive, existing states of affairs, not negative ones. Additionally, universal statements presuppose the existence of their subjects.
Chapter 15: What is the act of identification itself? “Identifying occurs whenever two representations are used together in a way whose success depends on their sharing the same real value.” Identifying “brings together manifestations of the same thing that have traveled through different causal routes.” Additionally, a concept is fundamentally an ability to identify something.
Chapter 16: A property is what it is because it stands in opposition to a range of contraries. A substance is what it is because it consistently selects one property from each relevant contrary range. Knowledge and reference are possible because these objective structures exist independently of thought and language.
Chapter 17: Persistence problems often concern how we collect temporal stages into wholes rather than anything about identity itself. Enduring objects possess subessences: stable, law-governed structures that preserve properties through time.
Chapter 18: How do we test whether a concept is adequate? The law of noncontradiction is the fundamental epistemic test for concepts because it mirrors the ontological structure of reality itself.
Chapter 19: Concepts originate as practical abilities for identifying stable features of the world, and they can be tested through repeated, coherent use in relatively small conceptual networks.
In conclusion, the author argues instead that meaning, reference, concepts, and knowledge are natural-world phenomena grounded in our practical abilities to identify and interact with things.
Chapter 2: Adapted Devices and Adapted and Derived Proper Functions: She defines relational proper functions, adapted proper functions, adapted devices, derived proper functions, etc. all conditioned behavior, whether relational or nonrelational, is behavior that has derived proper functions. Then she tries to explain how learned behavior can have functions–they are derived proper functions because they are the result of systems that were evolved to do things and so who have direct proper functions. Additionally, mental states are kinds of adapted devices and then human actions derive functions from them.
Chapter 3: Indicatives, Imperatives, and Gricean Intentions: She argues that one aspect of the meaning of a word, syntactic form, or other language device is its focused stabilizing function. The full direct proper function of any sentence is derived from the stabilizing functions of its significant elements or aspects, each of which belongs to its own reproductively established family. The stabilizing function of the imperative mood is to produce compliance; the stabilizing function of the indicative mood is to convey information, i.e., to produce true belief. Against Grice, It is very likely that small children speak and understand long before they have any such concepts of belief and intention.
Chapter 4: Language Device Types; Dictionary Senses; Stabilizing Proper Function as the First Aspect of Meaning: Denotative and referential terms are most special not because of their function but their intentionality which has more to do with the way a device Normally performs
its proper functions than with its proper functions per se. They function properly when an interpreter identifies or recognizes what their referents are. What makes two sentence tokens
tokens that display the same syntactic form should likewise be a matter of history (what reproductive family it is from), rather than mere physical form or actual function or a speaker’s intentions. What determines what a sentence meaning is history–stablizing function.
Chapter 5: Intentionality as a Natural Phenomenon: This is mainly an introduction to the main ideas she will discuss in the next section.
Chapter 6: Intentional Icons: Fregean Sense, Reference, and Real Value Introduced: She first defines intentional icons as including as being a member of a reproductively established family, stands midway between a Producer and Interpreter such that the interpreter can carry out its proper functions. What they represent is determined by the last member of the series of things it is supposed to map onto and the most proximate Normal explanations, respectively. Next, she responds to the problem of how words and sentences can correspond to nothing real. In response, she answers that what a sentence means is dependent on its history, of what its mapping looks like in Normal conditions. Then she argues that the "sense" of a language element is its Normal mapping rule. Importantly, intension is not the same as sense. Then she discusses real value which an intentional icon only has when it maps onto what it is supposed to. Lastly, when a term or token has a referent, there exists an x such that the term or token is supposed to map onto x.
Chapter 7: Kinds of Signs: Argues that intentionality comes in degrees using examples such as pressure gauges, adrenalin, art, and improvised signs.
Chapter 8: Hubots, Rumans, and Others: Case Studies of Intensions, Senses, and '"Stimulus Meanings": In this chapter, she argues that neither the "stimulus meaning" of a term nor its explicit intension (“rational being” -> “human”) determines its sense. The sense of a term determines neither its "stimulus meaning" nor its explicit intension. Senses—even the senses of thoughts—can be ambiguous (if the reason that an intentional icon is there is disjunctive) and they (or the purported thoughts) can be vacuous (if there is no relevant history). Additionally, lacking an explicit intension that has sense, a vacuous term is no more intentional than a sneeze.
Chapter 9: Intensions: Intension is meaning at all only insofar as intensions govern terms that
have sense or insofar as intensions are explicit and themselves have Sense. Additionally, the act of identifying the referent of a term or of coming to be able to identify it is not the same as the vague affair "coming to know what the term means."
Chapter 10: Simple Indexicals: An indexical does not have a fixed referent or fixed sense on its own. Instead, it has a relational sense: a rule that tells us how to find its referent from its relation to the context of use. This means they have proper functions which require contextual adaptors. Their referents emerge from real-world relations between speaker, hearer, and environment.
Chapter 11: Descriptions: Indefinite descriptions are normally descriptive and do not genuinely refer. Definite descriptions are primarily devices for introducing referents, not uniqueness.
The sense of a definite description consists in the mapping rules that determine which individual or property it picks out.
Chapter 12: Identity statements are for telling speaker to merge two concepts. It is not a relation. Existential statements tell the hearer to engage reference, and nonexistential statements the opposite.
Chapter 13: Explains typing, displaced function, and context-sensitive preservation rules in terms of what they’re supposed to do.
Chapter 14: Negative sentences correspond to positive, existing states of affairs, not negative ones. Additionally, universal statements presuppose the existence of their subjects.
Chapter 15: What is the act of identification itself? “Identifying occurs whenever two representations are used together in a way whose success depends on their sharing the same real value.” Identifying “brings together manifestations of the same thing that have traveled through different causal routes.” Additionally, a concept is fundamentally an ability to identify something.
Chapter 16: A property is what it is because it stands in opposition to a range of contraries. A substance is what it is because it consistently selects one property from each relevant contrary range. Knowledge and reference are possible because these objective structures exist independently of thought and language.
Chapter 17: Persistence problems often concern how we collect temporal stages into wholes rather than anything about identity itself. Enduring objects possess subessences: stable, law-governed structures that preserve properties through time.
Chapter 18: How do we test whether a concept is adequate? The law of noncontradiction is the fundamental epistemic test for concepts because it mirrors the ontological structure of reality itself.
Chapter 19: Concepts originate as practical abilities for identifying stable features of the world, and they can be tested through repeated, coherent use in relatively small conceptual networks.
In conclusion, the author argues instead that meaning, reference, concepts, and knowledge are natural-world phenomena grounded in our practical abilities to identify and interact with things.
January 28, 2024
This book (LTOBC) is so rich and dense. I don't know that I would have gotten as much out of it if I had not previously read the absolute masterpiece of pedagogy and synthetic exposition that is Beyond Concepts (BC), which makes frequent reference to LTOBC and allowed me to have the whole picture in mind from the get go. I went to read LTOBC attracted by the promise of a theory of natural functions, and boy was this delivered. But in the end, the master stroke, the thunderous proposition, is indeed the attack on what she calls "meaning rationalism", the idea that what our thoughts mean, or even only that our thoughts mean, that we have intentionality, is a priori or internal, something we can be sure of by inspection. This is such a huge deal that I won't know how to react to it for a while.
I would actually recommend starting with the more recent BC and then coming back to LTOBC if left wanting more, because it is a much more difficult book. But 5 stars for sure.
I would actually recommend starting with the more recent BC and then coming back to LTOBC if left wanting more, because it is a much more difficult book. But 5 stars for sure.
Want to Read
October 2, 2011Difficult book.
Worth the work?
Seems like it COULD be .. but ....
Worth the work?
Seems like it COULD be .. but ....
Displaying 1 - 6 of 6 reviews