Campbell Purton's website

HOME      Publications:      BUDDHISM        COUNSELLING       FOCUSING       PHILOSOPHY



Published in:  Philosophical Quarterly  Vol. 29 (1979), pp. 10-24







The term ‘function’ is one which is widely used by biologists. For example

we read in a textbook1 that


the function of light in photosynthesis is to supply the energy needed

for splitting of water ... and the function of chlorophyll is to trap

the energy and so make the splitting possible.        


It would also be quite appropriate to say that the function of chlorophyll

in plants is to enable plants to perform photosynthesis, that a function of

the heart is to circulate the blood, and that the function of the thumping

noise rabbits make with their hind legs is, as Darwin put it, “to signal to

their comrades”.2 Of course the notion of function is often employed in

biology without the word itself being used. For instance a biologist may

speak of the role which chloroplasts play in plant metabolism, or he may say

that certain substances in the blood plasma serve to clot the blood. Con-

versely, there are uses of the word ‘function’ in biology which do not seem

relevant to the concept i am interested in. In particular there is the sort

of case where it is said, for example, that the rate of starch formation in a

plant is a function of the light intensity. Biologists themselves are often

aware that their employment of the notion of function is in some ways

problematical; for instance Robert Hinde has recently devoted a paper to

‘The concept of function’ in ethology3. Amongst philosophers of biology

there have been several markedly divergent accounts of the concept in

recent years, and i shall review some of the main positions that have been

held before going on to give what i think is a more satisfactory account.



One account of biological function which has been popular with both

biologists and philosophers4 links the notion with that of survival.  But it

is clear that it cannot be simple survival of the individual organism that is

involved, for there are many organs, behaviour patterns and other items

which are said to have functions yet which do not contribute to the survival

of their possessors. There is the kangaroo’s pouch, the rabbit’s thumping

behaviour, the parental behaviour of many animals, and so on. If anything

such items tend to be detrimental to the survival chances of their possessor;

e.g., the rabbit’s thumping behaviour makes it more conspicuous to predators.

If we are to link function with survival chances it is presumably the survival

of the species (or of the relevant genes) which is important. Let us consider

the difficulties which any such account has to face.


First, an account of the function of an item in terms of what it contributes

to species survival leaves it obscure why we use the same term ‘function’

in contexts where there can be no reference to species survival. In particular

we speak of the functions of parts of machines, and here it is plausible to

say that a part has a function if it contributes to the purpose for which the

machine is used.  Thus in the case of machines it is human purposes which

underpin function ascriptions, and not any reference to species survival.

(If species had the purpose of surviving the problem would not arise, but

that there is any purpose behind organic development and reproduction is

a thesis which is not accepted by many biologists who effectively employ

the concept of function.) It could be held that we are simply dealing with

two different concepts here, one of which applies to the parts of machines,

the other to the parts of organisms, but I do not think that this is a very

plausible suggestion.        .


A second difficulty with the account which explains biological function

in terms of species survival is that an account of the concept of function

should not tied to a specific biological theory, even if that theory consti-

tutes the whole backdrop of biology, as Darwin’s theory does at present.

It should be possible to explain the term ‘function’ in a way that would be

acceptable to a Lamarckian or even an Aristotelian biologist. But if we ‘are

thus permitted to consider alternative biologies we can easily find examples

of functions which would be ruled out by the species-survival account. For

example an organ which helped an animal to satisfy its desires without

contributing to species survival could easily evolve in a Lamarckian world.

If an animal enjoyed getting a good view of its surroundings and as a result

its neck length increased slightly due to frequent stretching up, the increased

neck length would be passed on to its descendants, and a race of Lamarckian

giraffes would ultimately be produced of which it would be very natural to

say that the function of their long necks was to enable them to get a good

view of their surroundings.


Another pre-Darwinian example would be this:  Suppose the warning

thumps of rabbits alert field-mice, to the presence of danger. A modern

biologist would not recognize the thumping as having this function unless

the survival of field-mice in some way contributed to the survival of rabbits.

But judging from some remarks Aristotle makes about dolphins in the

Parts of Animals (696b 27) there can be little doubt that he would have

regarded the thumping as having this function. Such a view would be

entirely plausible within Aristotle’s scheme of things, because one would

see the thumping as contributing to the natural purpose of preserving the

species of field-mice.  Similarly, I think,  a seventeenth-century biologist

such as John Ray would see the warning of field-mice as one more example

of God’s marvellous design in the world, and would agree that one function

of the thumping was to warn field-mice of danger. But if these would be

quite proper uses of the term ‘function’ in biology, given the relevant back-

ground beliefs, they are uses which the species-survival analysis cannot

account for.                                        ,  .


Even without invoking alternative biologies one can imagine cases where

an organ would naturally be said to have a function even though it did not

contribute to the survival of the species. The following hypothetical case

is due to Richard Sorabji.6   Suppose that in a certain species of animal we

find a mechanism that outs out pain in circumstances where the animal has

suffered a fatal thrombosis. By hypothesis the animal will die soon after

the mechanism begins to operate, so the mechanism can be considered to

make no contribution to the survival of the species. Nevertheless, Sorabji

argues, we are inclined to say that the mechanism has the function of shutting

off pain. Sorabji’s general claim is that something has a function if it confers

a good on the animal, and that there are goods such as the avoidance of

pain which cannot be reduced to the good of species survival. This sort of

view has been elaborated more recently by Woodfield.7 Of course even if

Sorabji is right to say that the pain-eliminating mechanism has a function  -  and I think he is right -  it does not follow that his explanation of why we

speak of function in this case is correct. Bearing in mind the connection

between function and purpose we have noted in the case of machine parts,

and also in connection with Lamarckian, Aristotelian and pre-Darwinian

theistic  biology, it might be more plausible to suggest that what underpins

the function ascription in the case of Sorabji’s mechanism is not that,

avoidance of pain is a good for the animal, but that it is a settled purpose

of the animal, something the animal characteristically seeks. i shall say

more about the connection between the concepts of purpose and function






I want next to consider an intriguing account of the notion of function

which has been put forward by Larry Wright.8 Wright holds that the basic

idea involved in saying that item i has function F is that i is there because

it does r.  If i is there for some other reason, or if it is there merely by

accident, then i in contributing to f does not have the function F. To give

a couple of his examples: to say that the function of the sweep second hand

of a watch is to enable seconds to be read easily is, according to Wright, to

say that the hand is there because it enables seconds to be read easily. The

second hand may have other beneficial consequences, such as keeping the

watch dial free from dust, but these consequences are not its function be-

cause they do not explain why it is there. In the same way, says Wright,

to say that the function of the heart is to circulate the blood is to say that

hearts are there because they circulate the blood. According to Wright this

is true in two ways. First, a particular heart exists partly because of its

activity of pumping blood, without which the body, including the heart,

would have died. Secondly, in the species hearts have circulated blood in

the past and such circulation has contributed to the survival of the species.

Hence present-day hearts are there partly because hearts circulate the blood.

Either way, says Wright, one can say that hearts are there because they

circulate the blood.  The heart does many other things of course, such as

making heart sounds, but these other consequences of the heartbeat are not

functions of the heart because .they do not help to explain why the heart

is there.  To illustrate how “being there by accident” rules out the ascription

of function Wright gives the following example among others:


If [in a car engine] a small nut were to work itself loose and fall under

the valve-adjustment screw in such a way as to properly adjust a

poorly adjusted valve, it would make an accidental contribution to

the smooth running of that engine. We would never call the main-

tenance of proper valve functioning the function of the nut. If it got

the adjustment right it was just by accident. 9


Wright holds that what rules out the ascription of function in this case is

that the nut is not where it is because of what it does; it is there by accident.10

By contrast the fan belt does have a function since it is there because it

drives the fan.


Perhaps the most striking .thing about this account is that it contains no

references to either purposes or goods. Teleology seems to have vanished,

and to have been replaced by a rather curious form of causality. More

important, perhaps, is the point that .the, statement ‘i is there because it

does F” is ambiguous. It may mean ’i came to be there because it does F

or it may mean ‘i is kept there because it does F”. From most of Wright’s

examples one gets the impression that he favours the first alternative. He

holds that the sweep second hand of a watch has the function of enabling

seconds to be read easily even if the owner of the watch never uses it for

that purpose; that the function of a switch on the floor of a car is to dip the

headlights even if it is used only to scratch one’s foot, and so on.  Thus

what appears to be important is the consequence in virtue of which the

thing was put there and not the consequence in virtue of which it is maintained there (if it is).  Turning to the biological cases Wright holds that an organ

can properly be said to have a function if its coming to exist in the species

is due to its consequences for the organisms possessing it.  Wright puts it

like this:                ,

Organismic mutations are paradigmatically accidental. . . But that

only disqualifies an organ from functionhood for the first—or the

first few—generations.  If it survives by dint of doing something,

then that something becomes its function on this analysis. 11

It might seem at first sight that this case fits the ‘kept there’ interpretation

rather than the ‘came to be there’ interpretation. However, it can be con-

strued in terms of the latter in that the organ cannot be said to exist in

the species until after the first few generations. Thus it does in a sense come

to exist in the species as a result of what it docs. But while we may allow

Wright this case there is another sort of case which does not fit so well. An

organ which has function F1 in one environment may come to have a differ-

ent function F2 as a result of an environmental change. For example, the

swim-bladder of a lung-fish comes to function as a lung when the fish is out

of the water. Imagine a case where an organ comes into existence in the

species C2 is now its function. Here we cannot say that the organ exists

in the species (in the sense ‘has come to exist’) because it does C2. A similar

situation arises when an item exists in a species as a natural consequence

of the existence of items which do have a function. In changed environ-

mental circumstances such an item may acquire a function, but we cannot

say that having this function is a matter of its coming to be there because

of its consequences.


To avoid such difficulties it seems that Wright would have to opt for

the other interpretation of his thesis, i.e., that to say that something i has.

function F in system S is to say that i is maintained in S  because it does F.

Unfortunately this would let in items of which we would definitely not say

that they had a function.  Consider the following situation.  In a rocky

region a couple of more or less rectangular boulders stand embedded in the

ground a few feet apart. As a result of a minor landslip a similarly shaped

boulder falls on top of the two standing stones to form a table-like structure

resembling the megaliths of Stonehenge.  Call this third stone the cross-

piece.  As time goes on the soil is eroded from the base of the standing

stones so that but for the presence of the cross-piece they would fall inwards

and the structure would collapse. Now consider one of the standing stones.

It contributes to keeping the cross-piece up. But also it is there (maintained

there) because it keeps the cross-piece up; for if it did not do this the cross-

piece would not be where it is, and were it not where it is the standing stone

would fall. Any number of similar situations can be described in which

there are reciprocal causal relations between the parts of the system such

that one part is there (maintained there) because of its consequences for the

other parts. Counter-examples of this sort reveal, I think, the fundamental

weakness in Wright’s account of function, i.e., the lack of any reference to

purposes. Note that if I set up a three-stone structure in my garden (or if

I merely discover it there while cutting back the undergrowth12) and use it

as a garden table, one can say that the function of the standing stones is to

keep the cross-piece up.  But this is not because the standing stones stay

there because they keep the cross-piece up; it is because the structure has

acquired a purpose.




The idea that any adequate account of the notion of function requires

reference to the notion of purpose is not of course a new one. For instance,

Nagel holds that the statement that the function of chlorophyll in plants is

to enable plants to perform photosynthesis asserts the same as ‘A necessary

condition for the occurrence of photosynthesis in plants is the presence of

chlorophyll’, but he goes on to say that we only make functional statements

about goal-directed systems.” Now it is indeed clear that one cannot simply

analyse ‘i has function F in S’ by ‘i is necessary for F in S’. The existence

of the earth is necessary for the moon to move in its present orbit in the

solar system, but one would not say that a function of the earth was to

allow the moon to move in its present orbit. Nagel’s requirement that the

system under consideration must be goal-directed eliminates this sort of

counter-example. However, another sort of counter-example remains. Hem-

pel’s well-known example of heart sounds14 is a case in point.  The beating

of the heart in the body is necessary for producing heart sounds, and the

body is in Nagel’s sense a goal-directed system, but we would not say that

a function of the heart was to produce heart sounds. If we compare this

fact with the fact that it is a function of the heart to circulate the blood,

the difference appears to be that whereas the circulation of the blood contri-

butes to certain bodily ‘goals’ such as maintaining appropriate levels of

oxygen and carbon dioxide in the blood, supplying nutrients to the cells

and so on, the production of heart sounds does not seem to serve any ‘pur-

pose’. Thus we might usefully modify Nagel’s account by saying that i has

function F in S just if i is necessary for F in S, and F is or contributes to

a ‘goal’ of S.


Objections to this sort of account have centred round the necessity

condition and the goal condition. It has become clear, I think, that instead

of ‘i is necessary for F’ one needs a vaguer formulation such as ‘i does F

or ‘i contributes to F’.  The vagueness here parallels, or rather is an instance

of, the vagueness in ‘i causes F’, and I shall not discuss it further.  (Though

to avoid possible misunderstanding I should add that ‘does’ and ‘contributes’

are to be understood in a way that does not presuppose any sort of goal or

purpose. For instance producing certain sounds and vibrations will count

as something the heart does or contributes to.) Much more important are

the objections which can be raised in connection with the goal condition.


According to Nagel’s account, which is derived from an earlier analysis

by the biologist Gerd Sommerhoff, a system counts as goal-directed if its

overall state is specifiable in terms of several state variables subject to the

following conditions. It is assumed that the state variables are independent

of one another in the limited sense that the value of any variable at time t

is not determined by the values of any of the other variables at that same

time t.  (This condition rules out some equilibrium systems such as the

pendulum from the class of goal-directed systems.) Secondly the system

must be so constituted that it normally has a property G, and further that

if deviations from G occur due to some change in one of the state variables

then the values of the other state variables will alter, with the result that

the system returns to having property G. In view of what I shall say later

it should be noted that a Nagelian system need not be a system controlled

by negative feedback. In the latter sort of system it is deviations from G

which cause changes that in turn eliminate the deviations. In a Nagelian

system on the other hand the deviations from G may have no causal con-

sequences at all; G may be maintained simply because changes in one state

variable cause a particular sort of change in the other state variables. In

Kacser’s useful phrase15 Nagelian systems include systems which are merely

“buffered” against changes in G  rather than being “maintained” in state

G  in a strong sense. (Just what this “strong sense” involves I shall explain

more fully presently.)


One crucial difficulty with Nagel’s account is connected with the point

I have just been making. It is that the account applies to too many systems,

including some that no one would regard as goal-directed. Beckner16 points

out that according to the Sommerhoff-Nagel account an electrical resistor

carrying a current would count as goal-directed towards exemplifying Ohm’s

Law. Similarly, the meteorological system, in which the volume of water in

the sea remains at a more or less constant level despite evaporation, would

count as goal-directed for Nagel, but we do not speak of the function of

rain in this system.


A quite different sort of difficulty with Nagel’s account, at least if it is

taken as a general account of function, is that it does not apply to the

functions of the parts of artefacts. Of course if the artefact happens to be

a system of the sort Nagel describes then there is no problem; but most

artefacts are not Nagelian systems, yet we do wish to speak of the functions

of their parts. Given the difficulties inherent in Nagel’s account, it might

now be suggested that it is only in the case of artefacts that we have a work-

able concept of function. It might be held that parts of organisms can be

imagined as having functions through imagining that organisms were de-

signed for a purpose, but granted that they evolved through a process of

mutation and natural selection, it would have to be conceded that organic

parts do not really have functions. This sort of position is taken by Manser,

for example.17 Nevertheless, if we adopt this view we shall be committed

to saying some very odd things. A great many biological concepts, anatomical

and behavioural, are functional concepts. For instance, an eye is an organ

whose function is sight; a threat display, according to many ethologists, is

one whose function is to cause flight.18 Thus if we are to say that animal

parts and behaviour patterns do not really have functions it would seem to

follow that we must either say that animals do not really have eyes, or that

after all ‘eye’ is not a functional concept; but neither of these options seems

at all attractive. 




The position we have reached is this. We have a reasonably clear under-

standing of the concept of function insofar as it is applied to the parts of

things which have a purpose in the sense of being used for a purpose. In

what follows I shall refer to such purposes as “use-purposes” to distinguish

them from the “aim-purposes” (intentions, desires, etc.) of people and ani-

mals. Thus we can say that a thing has the use-purpose X just if it is used

for doing X.  On the other hand it seems that we employ the concept of

function also in biological contexts where the notion of a use-purpose is out

of place. Now the only straightforward way of reconciling these facts, I

think, is to show that while biologists do not employ the concept of a use-

purpose, they employ a notion which is sufficiently similar to it to allow the

employment of a concept of function. The notion which, I suggest, may fill

this role is the notion of a “maintained state” (or “maintained activity”)

to which I briefly referred in connection with Nagel’s account. By a “main-

tained state” of a thing I mean a state of the thing such that deviations

from that state result in changes which in turn eliminate the deviations.

Now there are two very different sorts of thing which can be said to have

maintained states (or activities) in this sense. First, anything which has a

use-purpose will have a maintained state or activity. For example, in the

case of a clock, so long as it is actually used for telling the time, it will be

maintained in such a way that, for instance, its larger hand rotates steadily

at a rate of about one revolution per hour. If this rate of rotation alters,

the clock will be adjusted, repaired, cleaned or otherwise attended to in

such a way that the deviations from the maintained activity (rotation at

one revolution per hour) are eliminated. Anything which has a use-purpose

must in its structure or behaviour satisfy certain conditions; certain states

or activities must be maintained in it.  These characteristic maintained

states or activities constitute the proper working conditions of the thing,

and it is with respect to them that we trace the functions of the parts of

the thing. For example, we say that the escapement ratchet of the clock

has the function of allowing the drive wheel to rotate slowly because that

contributes to the maintained activities of rotating the hands at their set

rates. To see the parts of the clock as having functions is precisely to see

them as contributing to the maintained activities of the clock.

The other kind of thing which can be said to have a maintained state or

activity is a thing such that a state or activity of that thing is regulated

by negative feedback. For example, in the circulatory system of the body

the concentration of carbon dioxide in the blood is maintained at a constant

level. Any significant deviation from that level will result in the stimulation

of nerves which transmit ‘impulses to the heart-rate centre in the brain,

which in turn will alter both heart rate and breathing rate, thus eliminating

the initial deviation. Here we have a system with a maintained state, and

the biologist is interested in understanding just how the elements of the

system operate so as to produce the maintained state. And, as in the case

of the clock, one identifies the functions of the. various elements through

discovering what it is that each element does by way of contributing to the

maintained state. In sum, my claim is that the notion of function gets its

primary application in the case of parts of machines and other things with

use-purposes. (I leave aside the question of whether this “primary” applica-

tion was in fact historically derived from the functions of individuals in

institutions.)  However, things with use-purposes have enough in common

with systems regulated by negative feedback (namely, both exhibit, main-

tained states) to allow us to extend the use of the term ‘function’ to the

parts of the latter kind of system. It would be wrong to think of this exten-

sion as introducing any ambiguity into the term ‘function’, since in each

case the force of questions about the function of a part is the same. In other

words function questions are questions about what exactly the part does

which, contributes to a maintained state of the thing in question.



My analysis of ‘i has function F in S’ could be stated as ‘i does F and

F contributes to or is a maintained state of S” (the formulation ‘contributes

to or is’ being designed to catch two slightly different sorts of case). How-

ever, this formula needs amplifying in the following respect.  It is to be

taken as read that the item i is construable as a part (or characteristic) of

the system S. I think we only speak of the function of a thing when that

thing counts as a part of a larger whole in which it has that function: the

board of directors has a function in a business organization; the peripheral

capillaries have a function in the thermo-regulatory system of the body; the

pistons have a function in a car engine, and so on. But if a wall is kept up

by someone’s leaning a heavy garden roller against it, I think that while

we might say that the roller functions as a support, or that the purpose of

the roller’s being there is to keep the wall up, we would not say that that was

its function.  In this case it is quite clear that the roller is not part of the

wall, whereas a buttress that held the wall up (whether by accident or

design) could reasonably be said to be part of the wall and could reasonably

be said to have the function of keeping the wall up. Of course whether

something counts as a part of a larger system is often disputable, but that

is quite consistent with my position so long as the borderline cases for parts

coincide with the borderline cases for functions. There are many different

sorts of case to be considered here.  Astronomers would certainly not say

that the sun had a function in the solar system; nor of course does the sun

have a function in the photo-synthetic system of a plant (though light does -

see the quotation with which I began this paper). However, one could

consider the “system” constituted by the sun and plant life on the earth,

and in this system the sun contributes to a maintained activity, namely

photosynthesis, and according to my account the sun could thus be said to

have a function in this system. I do not think this is really counter-intuitive -

one cannot say baldly that the sun has no function in the world of living

things.  Some other borderline cases which I cannot discuss in detail here

are the following:  Should we speak of the function of nitrogen-fixing bacteria

in the life of plants, or the function of the tick-bird in the symbiotic system

comprised of rhinoceros and tick-bird? Suffice it to say that in such cases

the hesitation we may have in speaking of function is probably due to the

hesitation we have in speaking of “parts”. In short, ‘function’ and ‘part’

go together: things which have functions are parts (or sometimes character-

istics) of things which have maintained states.  I suspect that this may be

the reason why we hesitate to say that the nut has a function in Wright’s

example quoted above. The nut is not clearly a part of the engine in which

it might otherwise be said to have a function.  By contrast, if someone had

soldered it on I think we should be less dubious about ascribing a function

to it, even if it was not soldered on in order to keep the valve adjusted.19




I want now to consider in more detail how my account of function works

in the case of biological examples. I have already indicated how it would

work in the case of the traditional example ‘the function of the heart is to

circulate the blood’: blood circulation—or rather circulation at a specific

rate depending on the circumstances—is clearly a maintained activity in

the body. Deviations from a particular rate, which is set by cell needs at

the time, produce changed which eliminate the deviations. Since the heart-

beat contributes to a maintained activity it is quite correct, according to

my account, for biologists to say that a function of the heart is to circulate

the blood. (As I shall explain shortly this is not the only reason one might

have for ascribing this function to the heart, but it is one good reason.) A

very similar account can be given of the other standard example ‘the func-

tion of chlorophyll in plants is photosynthesis’. The production of carbo-

hydrates by plants is a maintained activity the full details of which are still

not fully understood. But it is clear, for example, that the manufacture of

carbohydrates in a plant cell is regulated—any excess produced being trans-

ported elsewhere, often to the roots, for storage. Chlorophyll, in contributing

to a maintained activity of the plant, clearly has a function. A rather

different sort of case is provided by examples such as ‘the function of a

tiger’s claws is to enable it to catch and kill its prey’. Here I think we can

say that catching and killing are themselves maintained activities of tigers.

For if one tries to prevent a tiger from catching and killing it will make

efforts to overcome the obstacle in its way. If it has not caught and killed

for several days it will make up for this deficiency when the opportunity

arises, and so on.  Tigers are, in other words,  “goal-directed”  towards

catching and killing, and the claws are said to have a function because they

contribute to the realization of these “biological goals” or maintained



The notion of a “biological goal”, which I am understanding as simply

a maintained state or activity of a biological system, prompts the question

of whether individual and species survival should be construed, as they

often are, as biological goals. I shall return to the matter of species survival,

but so far as individual survival is concerned my account has the interesting

consequence that this is not a biological goal of the individual organism,

since one cannot say that deviations from surviving result in changes which

eliminate the deviations. If an organism deviates from surviving that is the

end of the matter. It is often said that organisms are goal-directed towards

survival, or that animals, as Woodfield puts it, “strive for survival”.20

However, I think that on reflection one can see that this is not the case.

What animals strive for are more specific things such as catching prey,

killing, eating, mating, etc. It is true that animals which did not have most

of these biological goals would not last long, and hence would not perpetuate

their kind; hence lit is no surprise that the biological goals which animals

have are such that as a result of having them the animals tend to survive.

But survival itself is not a biological goal of the organism; it is simply a

consequence of halving certain biological goals. It follows of course that my

account, which does relate functions to “biological goals”, is to be dis-

tinguished from those accounts which link function with ‘the biological goal

of the survival of the organism’.


So far as species survival is concerned it is again the case that deviations

from survival can hardly have consequences that eliminate the deviations.

Hence survival is no more a biological goal of the species than it is of the

individual. (This I think is consonant with the idea that there is no purpose

behind the evolutionary process.) Nevertheless, it turns out that there is a

close link between the notions of biological function and species survival

which is perhaps best explained by way of an example. A rabbit thumps

when predators are near. Experiments show that the thumping is a straight-

forward causal consequence of certain physiological states of the rabbit

associated with fear. For example, in the laboratory a rabbit thumps when

frightened irrespective of whether there are other rabbits around.  It is

hard to see how the thumping contributes to any maintained state of the

rabbit, but ethologists would certainly be inclined to say that the behaviour

has the function of warning other rabbits of the approach of danger. If we

are to understand the use of the term ‘function’ in this kind of case we must,

I think, introduce the idea of a maintained state of the species.  We need

now to think of the species as the system S, and I think there is no reason

in principle why we should not do so.21 Such a system, like any other, will

count as having a maintained state in so far as deviations from some of its

characteristics result in changes which in turn eliminate the deviations.

Now one might say that it is a characteristic of the rabbit species that

rabbits avoid foxes. Suppose that there is a deviation from this character-

istic in that some rabbits stop avoiding foxes due to some change in the

genetic make-up of those particular rabbits. The result will be that the

deviant rabbits will tend to die young, and hence will be less likely to

propagate their kind than the others.  Hence the deviant rabbits will

rapidly be eliminated from the population, and this can be expressed by

saying that in the species deviations from the characteristic of avoiding

foxes will result in changes that eliminate the deviations.


If this is right then avoidance of foxes is a maintained activity in the

species, and since thumping alerts other rabbits when foxes are near, and

since this contributes to the avoidance of foxes, it will be quite correct to

say, according to my account, that a function of the thumping behaviour

is to warn other rabbits.  In general, if item i in a species does F, and if

deviations from F would result in the deviant animals reproducing less

successfully it will follow, according to my account, that I has the function

F.   But to say this is .to say that if i has survival value in the species then

according to my account i has a function in the species. In this way my

account partially vindicates the species-survival account of biological func-

tion.  It only partially vindicates it, however, since not all biological func-

tions are functions in the species; some are functions in the individual

organism.  And again the vindication is only partial because to define

function in terms of species-survival makes it quite mysterious why we also

speak of function, in contexts where there is no question of speaking of

species-survival.  If having survival value confers a biological function on

an item, then any satisfactory account of function should explain why this

is so.               


In the last two paragraphs I have distinguished between maintained

states of species and maintained states of organisms. Consequently we need

to distinguish between the function an item has in the individual organism

and the function it has in the species.  For instance, the statement that

chlorophyll has the function of enabling plants to photosynthesize can be

interpreted in terms of the contribution chlorophyll makes to the regulated

metabolism of individual plants, as I sketched earlier; but it is clear that

chlorophyll also has a function in the species. Deviant plants which do not

photosynthesize, or do not photosynthesize adequately, will rapidly be

eliminated, so that photosynthesis counts as a maintained activity in the

species quite independently of the fact that it counts as a maintained activity

in the individual. But if photosynthesis counts as a maintained activity in

the species, then chlorophyll, which contributes to photosynthesis, has a

function in the species. Most, though not all, biological function statements

are ambiguous between function in the individual and function in the species,

and it is partly this ambiguity which has given rise to radically different

accounts of the notion of function in biology.




Before concluding I would like to make two related points, one mainly

philosophical, the other more biological. Philosophical discussions of func-

tion have often raised the question of whether specifying the function of

an item in any sense explains that item. For example, if we tell someone

that the function of the green colour of certain caterpillars is camouflage,

have we explained why they are green? The answer to this obviously de-

pends on what counts as an explanation, and I think we may say that to

explain something is to eliminate difficulties in understanding that thing.

For instance, on the whole we expect events to have causes, and in many

scientific contexts what this amounts to is that we expect that for any

event E it should be possible to discover previous events such that, given

known laws, the occurrence of. E can be deduced. Hence if we discover an

event E such that we cannot work out its causal antecedents this constitutes

a puzzle. There is an obstacle to our understanding of E, which is eliminated

by the discovery of the relevant laws which do, after all, allow E to be fitted

into the causal pattern. It is in this sort of way, I think, that relating E to

prior events and to laws constitutes .an explanation of E.  .Now in addition

to making assumptions about causality the biologist (for good reasons which

depend on his background theory, evolutionary or otherwise) approaches

the study of an organism with the assumption that organs and behaviour

patterns almost always have functions.  Hence if we discover an item i which is such that we cannot work out its function this constitutes a puzzle. There is an obstacle to our understanding of i which is removed by the discovery of the relevant maintained activity which after all allows i to be fitted into the functional pattern. In this way, by relating i to certain maintained activities we explain it.  It is sometimes said that specifying the function of an item does not constitute an explanation of the item, on the grounds that mere specification of function does not explain how the item came to

exist. But the answer to that is that functional explanations are not directed

towards relieving puzzlement about the origins of items; they are directed

towards relieving puzzlement about the consequences of items. If someone

is puzzled about how the caterpillars acquired their colour he will only be

satisfied by a causal explanation (either ontogenetic or phylogenetic).  But

not all puzzles in  biology are puzzles about causality; some are puzzles

about what an item can possibly do which contributes to a maintained state

or activity of the organism or species.


The related biological point is this. If one mistakenly thinks that ascrip-

tions of function in a species require knowledge of how an item came to exist

in the species, one will be inclined to think that claims about the functions

of items are inevitably rather speculative, for it usually is a speculative

business to work out how particular organs and behaviour patterns have

come to be the way they are. This point has been made by Tinbergen, who

associates function with survival value, but separates both from evolutionary

origin:               I

I have always been amazed, and I must admit annoyed as well, when

I meet among fellow zoologists with the implied or stated opinion that

the study of survival value must necessarily be guesswork; and that

exact experimentation on the problem is in principle not possible.

I am convinced that this is due to a confusion of the study of natural

selection with that of survival value. I agree that the selection

pressures which must be assumed to have moulded a species’ past

existence can never be subject to experimental proof, and must be

traced indirectly, I think we have to keep emphasizing that the

survival value of present-day species is just as much open to ex-

perimental enquiry as is the causation of behaviour.29


The conclusion I would like to draw from this is that conceptual con-

fusion in biology can have unfortunate consequences for experimental re-

search. Conversely, perhaps, one may hope that conceptual clarification of

fundamental concepts such as that of function may be of some scientific



University of St. Andrews




1 P. Weisz, Elements of Biology (New York, 1961), p. 209.


2 C. Darwin, The Expression of the Emotions in Man and the Animals (1872)


3  In G. Baerends, C. Beer & A. Manning (edd.). Function and Evolution of Behaviour

(Oxford, 1975).


4  See e.g. J. Canfield, “Teleological explanation in biology”, British J. Phil. Science,

14 (1963); M. Ruse, “Functional statements in biology”, Phil. Science, 38 (1971).


5 Compare Ray on the “more than ordinary Use in the Creation for such Insects as

are vastly numerous”: The Wisdom of God Manifested in the Works of Creation (1691).

Reprinted in part in Science and Religious Belief, ed. D. C. Goodman, (Bristol 1973),

p. 217.


6  In “Function”, Philosophical Quarterly, 14 (1964).


7  A. Woodfield, Teleology (Cambridge, 1976).


8  L. Wright, “Functions”, Philosophical. Review, 82 (1973), and Teleological Explanations (Berkeley, 1976). The account has a certain plausibility; for example in a review of  Wright’s book (Philosophical. Quarterly, 1978) Woodfield, who previously held a view similar to Sorabji’s, appears to have been converted to Wright’s view.

9  “Functions”, op.  cit., p. 162.                

  10  I shall suggest later an alternative reason why we hesitate to ascribe a function to the nut.

11  “Functions”. Op. cit., p. 165


12  In the period between his paper (op. cit.) and his book (op. cit., pp. 114-6) Wright seems to have become aware of the problems for his account posed by this sort of case. As I understand it his reply would be that although the standing stones were not put there because they keep the cross-piece up, they would have been put there to keep it up had they not been there already.  This seems implausible to me, but I do not have space to discuss it further.


      13  E. Nagel,  The Structure of Science (New York, 1961), oh. 12.


14 C. G. Hempel, “The logic of functional analysis”, in Aspects of Scientific Explana-

tion (New York. 1866).


15  Kacser, H “Some physico-chemical aspects of biological organization”, included

as an  appendix to G. H. Waddington’ The Strategy of the Genes (London, 1957). See also W. C. Wimsatt, “The concept of feedback”, Boston Studies in the Philosophy of Science, 8 (1971).

16    Beckner, The Biological Way of Thought (Berkeley, 1968), p. 138.


17  A. R. Manser, “Function and explanation”, Arist. Soc. Suppl. Vol. 48 (1973).


18  Functional classification in biology raises a whole set of problems which I cannot

discuss here. I have discussed the matter in connection with ethological concepts in a

paper “Ethological categories of behaviour and some consequences of their conflation”,

Animal Behaviour, 26 (1.978).           


19  However, another thing which may make us hesitate to say that the nut has a

function is that in the context of mass-produced things like car engines ‘function’ may

be taken as elliptical for ‘intended function’ (what it was put there to do rather .than

what it does), and being there by accident obviously bars the nut from having a function in this elliptical sense.  The blurring of the distinction between ‘function’ and ‘intended function’ is, I think, a serious defect in Wright’s analysis. 


20  A. Woodfield, “Darwin, teleology and taxonomy”.  Philosophy, 48(1873), p. 46.

21  Pace Woodfield, who writes that “Species . . . to be accurate . . . are not systems

at all” (op. cit.  p. 189). But it seems to me that talk of accuracy is out of place here

since surely falls into the category of terms that are technical yet vague.


22 For an interesting discussion of this “background assumption” see A. J. Cain,

“The perfection of animals”, in Viewpoints in Biology, Vol. 3 (ed. J. Carthy and B.

Duddington, London, 1964).


23 N. Tinbergen, “On aims and methods of ethology”,  Zeits. f. Tierpsychol., 20 (1963).


24  A previous version of this paper was read to the Scots  Philosophical Club in St. Andrews in May 1978. I am indebted to several of those present for their helpful comments.



Return to beginning of paper


HOME      Publications:      BUDDHISM        COUNSELLING       FOCUSING       PHILOSOPHY