ISSN 1135-1349
Issue #16 -- December 2005. Pp. 38-45
Dismantling the Straw Man: An Analysis of the Arguments of Hume and Berkeley Against Locke's Doctrine of Abstract Ideas
Copyright © by SORITES and Rhys McKinnon
Dismantling the Straw Man: An Analysis of the Arguments of Hume and Berkeley Against Locke's Doctrine of Abstract Ideas
Rhys McKinnon

Both Berkeley and Hume made an effort to respond to Locke's theory of abstract ideas: Berkeley vehemently rejected it, and Hume offered arguments in further support of Berkeley. Berkeley was quick to label Locke's product as an «abuse of language.»Foot note 1_1 Hume was not nearly as severe in the language that he used in his arguments in support of Berkeley against Locke, but believed that he was entirely in agreement with Berkeley's own arguments. However, there are grounds on which to question the strength of both Berkeley and Hume's attacks on Locke's account of abstract ideas: both Berkeley and Hume, possibly unbeknownst to them, reach some conclusions that agree with Locke. A journey through a close exegesis of Locke, and the subsequent attacks made by Berkeley and Hume, will lead to a conclusion that many of the attacks were merely of a straw manFoot note 1_2 nature and that many of their conclusions (Berkeley and Hume) can be taken to be in agreement Locke, instead of their believed stance of repudiating Locke. This paper is not a comparative analysis of the three philosophers' conceptions of abstract ideas; instead, it is an analysis of the attacks of Berkeley and Hume on Locke, given a close reading of Locke (as well as, obviously, Berkeley and Hume).

Locke is quick to state that humans have a need for abstract (general) ideas and terms: they allow for easier communication of ideasFoot note 1_3 Hume would certainly agree that the «custom» of abstract general ideas and terms facilitates communicationFoot note 1_4 but Berkeley certainly would not.Foot note 1_5 Locke argues this from the position that it is not very useful to attempt to communicate concepts with another using only particular cases, since it requires the other to be «acquainted with all those very particular Things» which are being used for referenceFoot note 1_6 Such a statement is what Aristotle concluded in his Metaphysics: experience is of particulars, but knowledge is of universalsFoot note 1_7 Although identifying the utility for abstract ideas is necessaryFoot note 1_8 (so that it is established as to why one should be concerned with the topic of the meaning and genealogy of abstract ideas), Locke's main focus is on the genealogy of abstract ideas/terms and what their specific meanings are. Locke makes a distinction between ideas and terms/words. Ideas are the objects of the mind when we thinkFoot note 1_9 and terms/words are the sounds/written symbols that we use to signify (use as signs) certain ideas in our mind.Foot note 1_10 Thus, ideas must be ontologically prior to terms, if the term is to have meaning qua being a sign for an idea.

Locke, as well as HumeFoot note 1_11 believes that ideas must come from either sensation (experience of the senses) or reflection (reasoning/«internal sense» of the mind.Foot note 1_12 In the genealogy of abstract ideas, and their general terms that signify the ideas, do the ideas come from sensation or reflection? Locke argues that there only exist particular things in the world, not general (abstract) things; thus, there cannot exist an object in the world from which the abstract idea can be arrived at strictly through the senses.Foot note 1_13 Therefore, if there are any abstract ideas, they must be from reflection. Ideas are made general (abstract) by separating from the particular idea all of the determinate qualities (viz. time, place, colour, size, etc.) that differentiate that specific idea from others that are similar to it (which one would group under the same abstract idea.Foot note 1_14 However, one cannot abstract all of the idea's qualities out of the particular instance: one must retain the defining qualities that make that idea of the sort that that idea belongs. An example will illustrate this last point. If one is to abstract from the idea of a specific triangle (a right triangle with vertical length of 3cm, horizontal length of 4cm, and hypotenuse length of 5cm), one can separate the ideas of determinate length and colour, but not of it being a «plane figure with three sides.» One cannot remove the last qualities since they are what make a triangle a triangle: they are its essential qualities. Locke defines abstract ideas in terms of these essences: every distinct abstract idea has a distinct essence;Foot note 1_15 but, he makes a distinction between supposed real essences, and nominal essences. His nominal essences are those in which we define the essence of a thing by a particular grouping of qualities which are defined by ourselves: such essences are nothing but «the uncertain and various Collection of simple Ideas» which our minds put together.Foot note 1_16 Real essences, conversely, are epistemologically inaccessible for us, and are the characteristics of its substance that makes it that particular thing and not another:Foot note 1_17 this is known as substance theory.Foot note 1_18 Interestingly, Locke's conception of the nominal essence is what Berkeley and Hume would call the essenceFoot note 1_19 of a thing: a bundle of perceptions that are constantly grouped together in our minds; this is known as bundle theory. In fact, Locke thinks that when we create general ideas, and terms for those ideas, we cannot be using the real essences of a thing. If we were to attempt to use the real essences, of which we have no knowledge, then we would not be able to precisely determine when a thing ceases to be a horse, and begins to be lead;Foot note 1_20 thus, to posit the existence of real, epistemologically inaccessible, essences «is so wholly useless» in our conception of abstract ideas.Foot note 1_21 However, if we define the abstract idea of a thing, viz. the abstract idea of horse, by using nominal essences, we are merely defining the thing (the abstract idea) by a group of simple ideas put together by the mind; in fact, Locke equates nominal essence («general natures») with abstract idea.Foot note 1_22

The main point of contention between Berkeley and Hume, on the one hand, and Locke, on the other, is the separability of ideas and qualities. In Hume's epistemology, he makes the bold claim «that there are not any two impressions which are perfectly inseparable.»Foot note 1_23 Thus, he claims that all things that are different are distinguishable, and all things that are distinguishable are separable by the mind (through the use of reason.Foot note 1_24 However, this is not entirely true: both Berkeley and Hume wonder if there are certain things (ideas or qualities) that are, strictly speaking, inseparable. The body of Berkeley's attack on Locke's doctrine of abstract ideas (and, consequently, Hume's furthering of Berkeley's arguments), is his argument for the inseparability of certain qualities. Since all three (Locke, Berkeley, and Hume) are arguing for bundle theoryFoot note 1_25 the pith of a given abstract idea is a collection of simple ideas: it itself is a complex idea (qua an amalgamation). However, it is in how Locke claims abstract ideas are made that the dispute lies: can we generate an idea of a thing without a definite value to each quality, viz. colour, size, motion? Berkeley and Hume, strictly speaking (loosely speaking, we will see that Hume offers another answer), do not believe that this is possible. Berkeley argues that there are certainly some parts or qualities that one can abstract from a given thing if and only if those parts or qualities can exist by themselves.Foot note 1_26 For example, it is possible for one to abstract the idea of nose from a man since it is readily conceivable that a nose can exist without the rest of the body.Foot note 1_27 Therefore, Berkeley argues that the idea of a Centaur (an example akin to Locke's unicorn and mermaid examples; Essay, 3.3.19) is intelligible to the mind, even though one may never have experienced one (as the mind alone is capable of putting the two ideas together in a non-contradictory way.Foot note 1_28 However, Berkeley and, subsequently, Hume believe that there are certain qualities that cannot exist apart from others; these include extension, colour, and motion. Berkeley argues that colour cannot exist on its own: it cannot exist without figure, which cannot exist without extension and motion.Foot note 1_29 However, Hume believes that ideas can be abstracted in the mind such that we can distinguish between different aspects of «inseparable» qualities.Foot note 1_30 To do this is to make a «distinction of reason».Foot note 1_31

A distinction of reason can be made when one considers two objects that are similar in one quality, but differ in (often all) others. Thus, the process is to discover different relations and resemblances of different sorts. Hume's example is the comparison of the following objects: a globe of black marble, a globe of white marble, and a cube of white marble.Foot note 1_32 If one were to merely observe and consider the white globe of marble, one would not be able to separate the colour from the figure; in fact, one would not even have a concept of the abstract idea of colour itself since one would only be acquainted with «a white colour dispos'd in a certain form.Foot note 1_33 However, if one were to consider the white globe in relation to the black globe, the differing ideas of colour have a resemblance (to each other) that can lead to the creation of an abstract idea of colour through this distinction of reason. Furthermore, if one were to consider all three objects together, one would notice that there are now two separate resemblances, viz. colour and figure/form, in what, at first, seemed to be single and inseparable.Foot note 1_34 To speak of the continuing inseparability, though we are able to reason an apparent separability through this distinction of reason, when we consider the white globe and its colour and figure, to do so requires a tacit reference to the black globe and white cube (or other objects that allow the distinction in reason through the two different sorts of resemblance, viz. that the objects only differ in one aspect from the original white globe: one with the same shape, but different colour; the other with the same colour, but different shape). Hume is here proffering the belief that simple ideas, though inseparable in reality, can have, in a sense, differing degrees of resemblance. This resemblance allows the two simple ideas of black and white to be grouped together and seen as -- only to a degree, since colour and figure are really inseparable -- closer in resemblance than between either (white and black) and roundness.Foot note 1_35 Finally, Hume believes that such a process gets better with practiceFoot note 1_36 and is, therefore, in agreement with the similar statement that Locke makes.Foot note 1_37 Berkeley, however, is in stark disagreement that the ability to abstract is developed since he does not seem to show a belief in the ability to make abstract ideas qua distinction of reason.

Thus, we have seen that Hume, while arguing in support of Berkeley, goes beyond Berkeley and, as a result, seems to be clearly agreeing with Locke in an important sense. Hume takes great care to show that although we can make a distinction of reason, there is no real separation of such qualities (one cannot truly separate colour from extension). This distinction is a point of departure between Locke and Hume: although Locke also argues for the real inseparability of ideas in the objects themselves, he still believes that qualities such as colour and figure can be separated from each other in the mind; however, Hume does not believe in the ability to really separate qualities in the mind. Thus, Hume, strictly speaking, does not believe that the qualities of colour and extension are actually being separated through a distinction of reason; it is only an illusion: to think of colour without extension is impossible.Foot note 1_38 However, we can see that there are different «sorts» of resemblances and, of these different sorts, colour and figure are members.

Finally, the other major point of contention between Berkeley and Hume, on the one hand, and Locke, on the other, is the nature of the abstract ideas themselves. As established above, Hume argues that we can have abstract ideas qua distinction of reason, but not abstract ideas qua qualities being actually separable (in the mind). In Locke's account of the genealogy of abstract ideas, as discussed above, abstract ideas are made by removingFoot note 1_39 all of the determinate qualities that differentiate a given particular thing from another, but leaving the qualities that the two particular things have in common as defined by the nominal essence for the abstract idea, of which these two things are members (all done in the mind, not in the things themselves). Thus, in the abstract idea of «human», the determinate qualities of size and colour, among others, are removed. Berkeley and Hume attack this contention. Berkeley argues as follows.Foot note 1_40 The abstract idea of man (human) must include all particular cases of humans. Since humans are of different colours, the abstract idea cannot have a determinate colour; the same applies to height, shape, and other such qualities. However, Berkeley cannot conceive of what such an idea would be since he cannot imagine an uncoloured person (more specifically, a person with indeterminate colour). Hume expands on this argument as follows.Foot note 1_41 Hume's argument is, in my opinion, a psychological account of what it is to consider an idea. Firstly, it is impossible for one's mind to conceive of any quality without having a precise degree. When one considers a line, it has a certain length; when one thinks of a triangle, it will have determinate lengths and angles thus defining its shape. Furthermore, when one thinks of a triangle, it must have both determinate shape and lengths; one cannot exist separated from the other (this is part of Hume's argument for the real inseparability of qualities even in the mind). Thus, this inability to conceive of a triangle without determinate qualities implies a contradiction to Locke's abstract idea of a triangle without any determinate qualities (other than the defining qualities of being a plane figure with three sides). Secondly, Hume argues that it is impossible for a triangle to exist which «has no precise proportion of sides and angles.»Foot note 1_42 Furthermore, it is impossible for an idea that has both quality and quantity but no precise degree in either.Foot note 1_43 Therefore, according to Hume's epistemology, viz. if one cannot have an impression of a thing then one cannot have an idea of it, we cannot have any abstract ideas qua Locke's conception of abstract ideas.Foot note 1_44 Interestingly, these arguments come from a fundamental misunderstanding of Locke; thus, a critique of these arguments will show that they are attacking points that Locke never made.

A close reading of Locke finds him explicitly stating that he does not believe that there are any existing general things.Foot note 1_45 As discussed earlier, Locke's epistemology distinguishes between ideas from sensation and reflection. Since there are no existing general things to allow the idea to come from sensation, the idea of abstract ideas must be from reflection.Foot note 1_46 The force of Berkeley's and, subsequently, Hume's attacks is usually taken to be their proof that there cannot be any existing general things. However, since Locke never claimed that any such things existed (he even also explicitly denies the possibility of their existence!), their arguments were merely of the straw man nature.

Earlier, it was implied that there may not be any abstract ideas. In order to address and resolve such a statement, one must first define what an abstract idea is. One can either take Locke's conception, which includes the perfect separability of qualities, or Hume's «distinction of reason.» Recall that Hume's distinction of reason is not merely our ability to distinguish between different ideas in our mind that are really inseparable in the objects themselves; instead, it is an illusion that we are making this separation in our minds: strictly speaking, we are not able to separate the ideas even in our minds. Since it is truly impossible to think of a colour without some extended body on which to imagine it, I find Locke's belief in the separability of colour from extension impossible. Thus, I find his conception of abstract ideas to be untenable. Hume's account, however, seems to fix this -- through his distinction of reason -- and has a more complete account of what abstract ideas are than Berkeley (by discussing our ability to use a distinction of reason to form a different sort of abstract idea than that of Locke's). The only point of departure between Hume and Locke, with respect to abstract ideas, is that Hume does not believe in the perfect separability of qualities (separated in the mind). Consequently, there seem to be abstract ideas under Hume's conception, but not Locke's.Foot note 1_47

In summation, a close reading of Locke's actual arguments has shown that not only have many of Berkeley and Hume's arguments been of a straw man nature, but, other than the one (major) point of departure (the inseparability of qualities), they seem to agree with Locke's account of the genealogy of abstract ideas, and their corresponding general terms (the terms are merely signs for the ideas).Foot note 1_48 Unfortunately, due to space constraints, this paper was not able to discuss the full extent of either Berkeley's or Hume's accounts on the genealogy of abstract ideas (Hume's argument from «custom» for example), but I do believe that the issues surrounding the actual strength of the attacks on Locke by these two prominent philosophers have been furthered.

Works Cited:

1. Dicker, G., Hume's Epistemology and Metaphysics. (London: Routledge, 1998).

2. McKeon, R., ed. The Basic Works of Aristotle. (New York: Random House, 2001).

3. Nidditch, P.H., ed. John Locke: An Essay Concerning Human Understanding (Oxford: Clarendon Press, 1975).

4. Norton, D.F. & Norton, M.J., ed. David Hume: A Treatise of Human Nature. (Oxford: Oxford University Press, 2002).

5. Woolhouse, R., ed. George Berkeley: Principles of Human Knowledge/Three Dialogues (London: Penguin Books, 1988).

Rhys McKinnon

University of Victoria

British Columbia, Canada

<rhys.mck at>

[Foot Note 1_1]

Berkeley, Principles of Human Knowledge, Introduction, 6; p38.

[Foot Note 1_2]

«Straw man» referring to the logical fallacy of attacking a position that someone never actually offered.

[Foot Note 1_3]

Locke, Essay Concerning Human Understanding, 3.3.3.

[Foot Note 1_4]

Treatise,, for example.

[Foot Note 1_5]

Principles, Introduction, 14.

[Foot Note 1_6]

Essay, 3.3.3.

[Foot Note 1_7]

Aristotle, Metaphysics, 1.2, 982a.

[Foot Note 1_8]

This is, certainly, not a conclusive argument, as Berkeley argues (see note13); but it may have weight if the particular (thing) that the other person is acquainted with is not sufficiently similar to that of the person who is offering their own particular (thing) in their discussion.

[Foot Note 1_9]

Essay, 1.1.8.

[Foot Note 1_10]

Ibid. 3.1.2.

[Foot Note 1_11]

Treatise, 1.2.1, though Hume makes an even more specific distinction between «ideas» and «impressions» over and above Locke's «loose» (in Hume's mind) use of the word «idea». Thus, in one sense, Hume would be discussing impressions, and another ideas; while Locke is only referring to ideas (as they have different definitions of what are ideas, and how they are made).

[Foot Note 1_12]

Essay, 2.1.2-4. Although, in the strictest possible sense, Locke and Hume differ in their accounts on what is meant by Sensation and Reflection, and how ideas are made/received, I believe that their respective meanings are certainly close enough to allow the equivocation of their two sentiments about this issue.

[Foot Note 1_13]

Ibid. 3.3.6, but more forcefully and explicitly in 3.3.11.

[Foot Note 1_14]

Ibid. 3.3.6.

[Foot Note 1_15]

Ibid. 3.3.14, 28.

[Foot Note 1_16]

Ibid. 3.3.14, 25-27.

[Foot Note 1_17]

Woolhouse, R.S., The Empiricists, 85-6.

[Foot Note 1_18]

Dicker, G., Hume's Epistemology and Metaphysics, 15.

[Foot Note 1_19]

Hume discusses this in terms of substance in 1.1.6. His definition of substance is first merely what Locke would call the nominal essence of a thing, but that this collection of qualities are somehow inseparably connected (by either contiguity or causation). Thus, I do not find it incorrect to use the word «essence» in the sense that I have in this sentence. Berkeley, however, never spoke in terms of either essence or material substance; thus, his inclusion in this sentence was only because of the way that he defined what it is to be a certain thing: to be a collection of a certain set of ideas.

[Foot Note 1_20]

Essay, 3.3.13.

[Foot Note 1_21]

Essay, 3.3.17, his emphasis.

[Foot Note 1_22]

Ibid. 3.3.9, lines 7-10.

[Foot Note 1_23]


[Foot Note 1_24]


[Foot Note 1_25]

Only when discussing abstract ideas. Locke argues for substance theory with respect to material objects, where Berkeley argues for bundle theory. Hume, however, neither commits himself to bundle theory nor substance theory: he is skeptical about substance theory, but does not want to dismiss its truth. Thus, he argues for bundle theory as the only epistemologically tenable theory, given our current state of epistemology, but not explicitly for the «truth» of bundle theory.

[Foot Note 1_26]

Principles, Introduction, 10.

[Foot Note 1_27]

I say this keeping in mind that some philosophers, Aristotle comes to mind, would argue that this conception is not of a nose qua nose because the idea of a nose is only when it is functioning as a nose (therefore it must still be attached to the body). However, it is neither inconceivable nor contradictory to imagine a nose removed from a man, even if we should call it a name other than «nose.»

[Foot Note 1_28]

Ibid. 10.

[Foot Note 1_29]

Ibid. 7.

[Foot Note 1_30]


[Foot Note 1_31]


[Foot Note 1_32]

Ibid. I believe that this example is best taken if we consider these three objects as our only experiences of objects.

[Foot Note 1_33]

Loc. cit. Although it is interesting that he is using language that is meaningless if we only have the experience of this one object.

[Foot Note 1_34]

Loc. cit. It is important to note that although these qualities now appear separable, they still, in reality, are not (according to Hume).

[Foot Note 1_35]

This is an expansion of part of the Appendix to volume 3 but was inserted into Book 1 found on page 18 of Norton, 2002.

[Foot Note 1_36]


[Foot Note 1_37]

Essay, 4.7.9.

[Foot Note 1_38]


[Foot Note 1_39]

I have here, and hereafter, used «remove» (and its cognates) to describe Locke's genealogy of abstract ideas. This is a result of a semantical interpretation of Essay, 3.3.8. He speaks in terms of: «retaining only those Qualities», «uniting them», «leaving out», and not creating new abstract ideas by «addition.» Thus, I take «leaving out the shape, and some other Properties» and «retaining» a specific set of others to mean «removing» the qualities not wished to be «retained» in the abstract idea. The use of «remove» seems to be consistent with the overall meaning of this section (3.3.8).

[Foot Note 1_40]

Principles, Introduction, 9-10.

[Foot Note 1_41]

Treatise, The most pertinent sections for our current discussion is 2, 3, and 6.

[Foot Note 1_42]


[Foot Note 1_43]

Loc. cit.

[Foot Note 1_44]

It is of paramount importance to remember that this is only Hume's argument, not a critique of it.

[Foot Note 1_45]

Essay, 3.3.11. See note 10 as well.

[Foot Note 1_46]

By using the disjunctive syllogism (process of elimination), see note 20.

[Foot Note 1_47]

Since I found Locke's definition of abstract ideas, viz. including the perfect separability of qualities, to be untenable.

[Foot Note 1_48]

I say «they» because Hume is, in his mind, arguing in favour of Berkeley; however, Berkeley may have also rejected Hume's account of abstract ideas qua distinction of reason (but this is only conjecture since Berkeley never explicitly brings it up).