Wittgenstein's Theory of "Family Resemblance" Concepts

Submitted for IB Extended Essay (Philosophy)

Introduction

Ludwig Wittgenstein is a philosopher of unparalleled notoriety – both for his deft logical precision, and his tremendous force of character. This is a man both with the ego to believe he had solved all the problems of philosophy, and the intellectual humility to return and say he was wrong. Indeed, the extent of this turnaround was so great that Wittgenstein’s thought is often divided into two categories: the “Early Wittgenstein” and the “Late Wittgenstein”. The subject of this essay is an idea he had in the latter stage of his philosophy: the theory of “family resemblance”.

In this essay, I will begin by identifying its place within both Wittgenstein’s thought and in the broader context of the Western philosophical tradition – the sharp contrast against both of these backdrops is a useful aid to understanding it. Then, I will elucidate the theory with particular reference to his Philosophical Investigations. I will seek to determine and defend Wittgenstein’s thought process in expounding this theory, taking into account criticisms from subsequent philosophers. Finally, I will conclude that although Wittgenstein commits no fallacy in constructing his theory, it is susceptible to being invoked in a fallacious manner. That is, it is a theory of how words are defined in natural language; it cannot be imported into systems in which we alter the way language is used, such as in science.

What is the theory of “family resemblance” concepts?

I will begin with an explanation of what is meant by the theory of family resemblance. In the preface to the Philosophical Investigations, Wittgenstein holds that these ideas “could be seen in the right light only by contrast with and against the background of my old way of thinking” (Wittgenstein, Philosophical Investigations, 1953, p. x). These ideas are not only a reaction his earlier philosophy, but also to the Western philosophical canon as a whole. Hence I will explicate the theory in two lights: a response to Plato, and a response to the earlier Wittgenstein.

Family Resemblance as a Response to Plato

In Plato’s dialogues, Socrates often attempts to define some important abstract concept. In the Meno, for instance, Socrates and Meno discuss virtue, and whether it is taught, or the result of practice, or simply in the nature of some men (Plato, 1976, p. 70a). To answer this, they agree that they should first define what is meant by virtue. Socrates claims to not know what virtue is, adding, “Not only that […] I have never yet met anyone else who did know.” (Ibid., p. 71c)

When Meno proffers his definition of virtue, it quickly falls apart under Socrates’ questioning. Eventually, he concedes the argument, and Socrates asks, “Shall we try to find out together what virtue is?” (Ibid., p.86c) – and this is the objective of the remainder of the dialogue.

Socrates clearly has an ability to relentlessly probe the definitions of others, extract inconsistencies, and confuse his opponents into submission. Though Socrates’ contemporaries have an understanding of virtue that is actionable and practical, Socrates finds flaws in the articulation of that understanding.

A similar pattern of argument is seen throughout many of the other dialogues; characteristic of the 'Socratic Method'. Just as he dismantles definitions for virtue in the Meno, he tears down definitions of justice in the Republic, and definitions of love in the Phaedrus.

The premise is that we can define virtue, justice and love in terms of their necessary and sufficient qualities. This means that just as mathematicians can say a number is even if and only if it can be divided by two and leave no remainder, we should be able to define everything else in this way. Socrates effectively wins arguments by pointing out that nobody can tell him what quality all instantiations, and only instantiations of virtue have.

Wittgenstein’s rejoinder is therefore easy to understand: it is a straightforward rejection of this premise. As he says in the Philosophical Investigations, “Instead of producing something common […] I am saying that these phenomena have no one thing in common which makes us use the same word for all, but they are related to one another in many different ways.” (Wittgenstein, Philosophical Investigations, 1953, p. 27).

He gives the example of a game. We are asked to consider all those things which we call games: board games like chess; ball games like football; kids’ games like peekaboo. It is a mistake, claims Wittgenstein – to dogmatically insist that there must be “something common, or they would not be called ‘games’”. (Ibid.)

In other words, a thing is not beautiful just because it possesses some quality – a Platonic form of beauty – which is shared by all beautiful things, and only beautiful things. The relationship between beautiful things is not that they share in some essence. Instead, one beautiful thing may share some characteristic with another, and some other characteristic with another; and the two may not share any obvious characteristics with each other, except that they both resemble the first – just as member X of a family might have similar hair to member Y, who in turn has a similar nose to member Z. Members X and Z need not have features in common: they are noticeably related – of the same family – because they both resemble Y.

Family Resemblance as a Response to the Earlier Wittgenstein

Recalling Wittgenstein’s insistence in the preface of the Philosophical Investigations that his later ideas should be considered with reference to his earlier works, I will now turn to examining this relationship.

Wittgenstein was originally trained as an aeronautical engineer. In his studies, he developed an obsession with mathematics, and then with the foundations of mathematics – it was this fixation that eventually led him to Bertrand Russell in Cambridge, where he began philosophizing in earnest (Monk, 1990, pp. 30-40).

Given his background in mathematics, and the explicit contrast I have drawn between mathematical definitions and definitions by means of a family resemblance relation, one might wonder how Wittgenstein arrives at his theory. Unravelling this process shows it to not be a simple, common-sense response to an unreasonable demand for rigour, but a sophisticated, self-contained theory in the philosophy of language.

Wittgenstein had published only one book prior to the Investigations: the Tractatus Logico-Philosophicus (henceforth, the Tractatus). Therein, he attempts to expunge dogma from philosophy. It delineates what can and cannot be put into a proposition, and examines all the entailments and implications of his views.

He concludes that “what we cannot speak about, we must pass over in silence.” (Wittgenstein, Tractatus Logico-Philosophicus, 1922, p. 74). He believes that he has determined the boundaries of meaningful speech. Discourse about topics such as ethics, religion and free will are beyond the perimeter of what can be known, and cannot be meaningfully spoken about. All potential philosophical problems that can arise are resolved simply because the questions arise in language, and he has built a framework of what can be meaningfully expressed therein; and once we reach the limits of our language, we have reached the limits of our thought. We cannot get beyond this, and a proposition that runs into this issue is simply meaningless (Sugrue, 2020).

With this triumph, Wittgenstein retired from philosophy. However, as the popularity of the Tractatus grew, so did the number of criticisms which he found trying. For instance, in response to Wittgenstein’s insistence that a proposition and its subject matter must share a logical form, his friend – economist Piero Sraffa – simply made a rude gesture, retorting, “What is the logical form of that?” (Monk, 1990, p. 260)

Thus, Wittgenstein revised his perspective. He had made the assumption that the task of logical analysis was to discover elementary propositions, whose form was not yet known. Concluding that he had been hypocritical, he transitions to the “Later Wittgenstein” – the Wittgenstein of the Philosophical Investigations. This is why the focus shifts from formal logic to the grammar of ordinary language (Stanford Encyclopedia of Philosophy, 2021).

Philosophy, he says, is “a battle against the bewitchment of our intelligence by means of language” (Wittgenstein, Philosophical Investigations, 1953, p. 40). To win this battle, he has utterly rejected the dogma he had previously held about language – and turns instead to family resemblances as a theory of how language is actually used.

Is the theory of family resemblance a fallacy?

Having delineated the theory, I will now turn to evaluating whether Wittgenstein does indeed commit a fallacy. The biggest criticism that I will deal with is that the scope of the theory is too extensive, since all things resemble each other in some manner or the other. I will argue that the theory of family resemblance is a description of a subset of language, and outline when it is a valid one. However, it cannot be applied to all language, and attempts to do so are certainly fallacious.

Let us begin with the immediate intuitive appeal of the idea. Consider the simple question of what art is, and what pieces qualify as works of art. This is the sort of quandary which Plato might use to claim that nobody truly understands what art is (else there would be no disagreement on what constitutes art); and then conjure a transcendental realm in which resides the form of art. Wittgenstein, by contrast, offers us a way to make sense of the question. If we accept that art is a family resemblance concept, then we do away wholly with the unnecessary search for a definition.

Indeed, it is not even limited to abstract concepts. Even mundane objects – beds, chairs, tables – do not have a set of necessary and sufficient qualities. As such, family resemblance is a straightforward way to obviate some of the inconsequential questions that might arise here – like whether or not a throne is a chair.

This is the sort of thing meant by ‘the bewitchment of our intelligence by means of language’. We misuse language in asking these sorts of things: whether a throne is a chair clearly depends upon the context in which it is being discussed. Wittgenstein calls these “language-games”, which he refers to throughout the Philosophical Investigations (Wittgenstein, Philosophical Investigations, 1953, p. 4). Indeed, the word ‘throne’ has no meaning whatsoever outside of its use in some particular language-game, he argues; its meaning could be an order, a question, or a reference to some particular throne depending on what language-game is being played. Similarly, the question ‘what time is it on the sun?’ is not a profound philosophical question because it defies an answer; it merely uses the word ‘time’ in a language-game that renders its meaning nonsensical.

Let us consider the Sorites Paradox. The ‘paradox’ is about the inherent vagueness of certain predicates – for example, we cannot say how tall one must be to be considered tall, since it is not defined in the form, “one is tall ‘if and only if’ one is height X or taller”. It is associated with the question of how many grains of sand must be piled together to qualify them as a heap. Since one grain of sand is not a heap, and adding a single grain of sand to a collection that is not a heap is not sufficient to make it a heap, it follows that we can never form a heap; however, heaps exist, hence the apparent paradox. (Allan, 2009, p. 1037).

Wittgenstein’s ‘anti-systematisation’ enables us to escape such paradoxes: we respond that a heap is a family resemblance concept. Although it is indeed a vague term that lacks the logical rigour sought by philosophers like Plato, it is an operable term in daily life. Heaps are defined by their resemblance to other heaps, not an ‘if and only if’ clause according to the number of grains or anything else.

However, this response immediately raises two questions. The first is concerned with the scope: we may well say that heaps are a family resemblance concept; perhaps other vague predicates associated with the Sorites Paradox like ‘tall’ and ‘bald’ are too. We have also illustrated it with chairs and tables, as well as abstractions like ‘love’ and ‘virtue’. Therefore we might wonder how we determine precisely which ideas are family resemblance concepts. It is not clear whether the claim is that it applies universally or only to some notions, and if only some, which ones. The second question is what precisely we mean by the word ‘resemble’, and how we determine whether something resembles another thing, and the necessary extent of the resemblance required to contend two things are the same. I will elaborate upon both of these criticisms, starting with the latter.

Let us return to the example of art, and to the question of what pieces constitute works of art. Evidently, there is no consensus for a definition that we could utilise to easily determine if some particular thing is art: no way of saying, “A thing is a piece of art if and only if it meets some particular criteria; if these are not met, then it is not”.

We have thus framed family resemblance as a way of obviating a search for a definition of this form, by claiming that it is a misguided way of understanding art. Instead, art can be identified as it resembles some paradigm, which everyone agrees is art.

The issue is clearly that this seems to give us valid reason to believe that anything that anyone can construe to be art, is art. Suppose that some famed artist puts up a broom and a bucket in an art gallery. This can of course be viewed as art, since it shares a common feature – has a family resemblance – to other pieces of art, namely that it is intended to be a work of art (and perhaps communicates some theme too). However, now consider a broom and bucket left in the same art gallery by a janitor. This resembles the other in its appearance, and in its location; so it seems Wittgenstein offers us scope to argue that the broom and bucket in this case are art – though our common-sense indicates that they are clearly not.

This is the criticism employed by Leon Pompa: “If we entertain the plausible hypothesis that between any two people there will always be some resemblance or the other, be it only that they each resemble some third person, it would thus become the case that we were all members of one family. And in this case the term ‘family’ would become vacuous, since it excluded nobody.” (Pompa, 1967, p. 65)

Pompa’s criticism, as I illustrated with reference to the example of art, is effectively that in Wittgenstein’s commitment to avoiding a definition, his account is too inclusive – all things resemble each other in some respect or the other.

Moreover, any attempt to restrict what it means for something to resemble another will eventually converge on a definition, precisely of the sort Wittgenstein is trying to avoid: if we said of the broom and bucket that to be art, they must resemble a paradigm of art in some specific respect, then resemblance in this respect becomes a necessary and sufficient condition for being works of art (Stanford Encyclopedia of Philosophy, 2018).

Of course, the criticism is equally damaging in other cases. Consider the more mundane example of dogs and cats: they resemble each other in being common household pets; in having four legs; and in being territorial. Similarities are not difficult to come by, and yet, clearly dogs and cats are distinct in the way we ordinarily use language.

This also raises the question of the initial example which we simply ‘recognise’ to be art (or whatever we are seeking to define). It is all well comparing the respects in which various things resemble some paradigm case, but this raises questions of how we decide which work or works fall into this category.

Hence, modern philosophers defend family resemblance by framing it subtly different to how I have treated it so far. Instead of arguing that for something to fall into some particular category, it must resemble something that is clearly a member of that category, some argue that we can proffer a list of properties, none of which is necessary, but which can be jointly sufficient for membership of that family.

I will continue to illustrate this with reference to art, though of course the claim is that this method holds more widely. Berys Gaut provides a list in this sense: possessing positive aesthetic properties; being expressive of emotion; being intellectually challenging; being formally complex and coherent; having the capacity to convey complex meanings; exhibiting an individual point of view; being original; being an artifact or performance which is the product of a high degree of skill; belonging to an established art form; being the product of an intention to make a work of art (Gaut, 2000, pp. 25-44).

This escapes the problem of justifying a paradigm case, by specifying resemblance in reference to general properties rather than particulars. However, it raises new issues. For instance, the same concept can have many lists of criteria, all of which appear sensible. It seems that the criteria we pick are somewhat arbitrary – how are we to choose which list to adopt? Further, it is not obvious that this conception of family resemblance succeeds in avoiding a definition: with such arbitrary and yet specific criteria, it appears to simply be a long, complicated form of the definition Wittgenstein seeks to avoid – “A thing can be said to be art, if and only if it fulfils some of the following criteria: …”

Upon close inspection, it seems to run into precisely the same problem as our original framing of family resemblance. Returning to the dogs and the cats, the ‘list’ of general criteria as to whether something is a dog or a cat must have overlap: an anatomy designed for four legs; carnivorous; etc.. The degree of overlap necessarily means that all dogs fulfil many of the criteria to be considered a cat, and vice-versa, though clearly they are mutually exclusive (one cannot simultaneously be a dog and a cat).

To salvage the theory however, Wittgenstein might argue that this ‘cluster’ conception of family resemblance is clearly more appropriate with reference to things like art, beauty, virtue or justice than it is for dogs. Thus we turn to the second of my major criticisms: the lack of clarity over precisely which ideas are family resemblance concepts.

Clearly mathematical definitions are not; and yet the aforementioned ‘abstract’ qualities (in which Plato is typically interested in) seem to be best described by family resemblance. My argument will be that family resemblance is a successful theory in ‘natural language’ – that is, when we place no artificial limits on speech – this explains its intuitive appeal. However, some systems of language, like science and mathematics, place limits on language. Therefore the same word can be used in two different language-games (natural and ‘unnatural’ language), and defined in a different manner in each. As a result, family resemblance does not apply to dogs and cats, because their definition is understood biologically – so appealing to family resemblance is a fallacy here.

The argument is perhaps best illustrated with reference to the question, ‘What is a woman?’. This was the title of a documentary concerning gender and transgender issues, because it has become fashionable in certain conservative circles: the premise is that the answer ought to be obvious, and yet the purported insanity of the American left has left many confused. The producer, Matt Walsh, notes that “[he] would have given an answer that is biological, because that is 100% the answer.” (Graham, 2022)

Walsh’s view is that the definition of a woman pertains to some set of necessary and sufficient conditions. It is a fundamentally biological question: a person is a woman if and only if some particular traits – for instance, two X chromosomes – are present.

Though this may have been the consensus view for decades, it is flawed. Some individuals feel body dysmorphia such that their gender identity does not correspond to their biological characteristics. Additionally, genetic defects exist which mean that people whom we would otherwise recognise as women fail to fall under a definition of this sort – for example, intersex.

A successful description of how language operates here needs to account for these: defining ‘woman’ in such a way that transgender women are recognised as such. If we are to extend our definition in this way, a Platonic definition is clearly inappropriate. One might say, “some person is a woman if and only if that person identifies as a woman”, but this is clearly problematic. Suppose a reactionary male pundit declares himself to be a woman; this framework would entail that we take their sardonic protest seriously, and allow them to compete alongside women in competitive sport. Further, it means that scientists lose rigour in studying distinctions between men and women.

Wittgenstein’s argument strikes a compromise that appears to solve the problem. Suppose that gender is a family resemblance concept. Thus we can extend our definition of ‘gender’ and ‘woman’ to fit the way they are used in ordinary language. We simply construct a set of criteria – say, a pair of X chromosomes; a sense of personal identity commensurate with being a woman; the ability to become pregnant before menopause; displaying traditionally feminine behaviours; and so on. An individual can be reasonably identified to be a woman if some or more of these criteria are met.

Feminist philosopher Ann Garry concurs: “… there is nothing – neither a property, an experience, nor an interest – that all women have in common, we know what a woman is and who women are because of crisscrossing, overlapping characteristics that are clear within social contexts.” (Garry, 2011)

The confusion arises because the question of defining a woman arises in two distinct language-games – in the cultural sphere, and in science. It is perfectly reasonable to assert that in day-to-day life, we do not identify whether someone is a woman through analysis of their chromosomes or genitalia. Instead, we see a set of various attributes, which cumulatively tend to give a strong indication.

In science however, language is used differently. The scientist seeks to evaluate causality – to isolate independent variables, and see what effect arises from varying them; and then to form hypotheses and predictions based on the result. This is why biologists and psychologists can point at differences that exist between men and women, be it neurological or physical. It is not done in the same vein as a cultural commentator noting the fact, but reliant upon individuals who can clearly and rigorously be identified as women, and further analysis of what this entails, and how it might or might not be different from a man.

This point – that family resemblance cannot be upheld in disciplines which restrict the way we use language naturally – holds more broadly. Science does this: it does not understand objects ‘intuitively’. Therefore, a fallacy arises in using family resemblance in this particular ‘language-game’, because it is only applicable with respect to natural language.

Another example of a field in which family resemblance does not apply is Law. It requires clear and precise definitions, in a manner that is distinct from how we would naturally use language. It would be a fallacy to argue, “execution ought to be illegal, because it is murder, and murder is illegal”. In natural language, some might well hold the view that execution is murder. However, the above argument nonetheless fails because it conflates two uses of the word ‘murder’: as an ambiguous term in natural language, and as a legal term. In the latter, a killing is murder if and only if it is unlawful (among other things).

Conclusion

I have demonstrated that the theory of family resemblance is sound, subject to caveats. It is an observation about the natural employment of language, not a systematisation intended to formalise the use of all language. Indeed, such systematisations are precisely what Wittgenstein is seeking to circumvent. It fails in those systems in which language is used artificially – I have shown that this occurs in science, maths, and law. As such, we must take care when the same word is used in two language-games, and the nature of its definition is distinct in each. Therefore it is fallacious to attempt to claim ‘woman’ is a family resemblance concept, and then argue about the scientific implications. The same goes for precisely defined legal terminology like ‘murder’ (or ‘fraud’).

Hence family resemblance in its own right is not fallacious, and succeeds as a description of natural language. Since this, rather than systematisation, is its explicit goal, it is philosophically sound.

Bibliography

Allan, K. (2009). Concise Encyclopedia of Semantics. Elsevier Ltd.

Garry, A. (2011). Intersectionality, Metaphors, and the Multiplicity of Gender. Hypatia, 838-839.

Gaut, B. (2000). Art as a Cluster Concept. In N. C. (ed.), Theories of Art Today. Madison: University of Wisconsin Press.

Graham, J. (2022, June 5). Podcaster Matt Walsh, The Daily Wire ask 'What is a woman? | Opinion - Deseret News. Retrieved from Deseret News: https://www.deseret.com/2022/6/4/23153551/he-asked-what-is-a-woman-most-people-couldnt-answer-matt-walsh-daily-wire-gender-ideology

Monk, R. (1990). Ludwig Wittgenstein: The Duty of Genius. Rushden: Random House.

Plato. (1976). Meno. Hackett Publishing Company Inc.

Pompa, L. (1967). Family Resemblance. The Philosophical Quarterly, 65.

Stanford Encyclopedia of Philosophy. (2018, August 14). The Definition of Art (Stanford Encyclopedia of Philosophy). Retrieved from Stanford Encyclopedia of Philosophy: https://plato.stanford.edu/entries/art-definition/

Stanford Encyclopedia of Philosophy. (2021, October 20). Ludwig Wittgenstein (Stanford Encyclopedia of Philosophy). Retrieved from Stanford Encyclopedia of Philosophy: https://plato.stanford.edu/entries/wittgenstein/

Stanford Encyclopedia of Philosophy. (2022, May 26). Socrates (Stanford Encyclopedia of Philosophy). Retrieved from Stanford Encyclopedia of Philosophy: https://plato.stanford.edu/entries/socrates/

Sugrue, M. (2020, July 20). Great Minds - Part 5 - The Latter Wittgenstein: The Philosophy of Language. Retrieved from

Wittgenstein, L. (1922). Tractatus Logico-Philosophicus. Routledge & Kegan Paul Ltd.

Wittgenstein, L. (1953). Philosophical Investigations. (G. Anscombe, Trans.) Blackwell Publishing Ltd.

Result

Mark: 26/34 (B)