What is the causal theory of perception? Is it true?
Submitted for PH144 (Mind and Reality) Summer Exam
![](/img/missing-image.png)
I shall explain the causal theory of perception, and argue that Grice’s argument for its truth prevails over criticisms from Smith, Jones and Snowdon. In particular, having delineated the argument and criticisms, I shall argue in favour of the sense-datum theory of perception and therefore for the common kind assumption that veridical and hallucinatory experience is of the same nature (Snowdon’s criticism being that Grice’s argument depends on this assumption). This being the case, I conclude that veridical perception requires an appropriate causal relation between object and perceiver, as the theory requires.
The causal theory of perception (henceforth, CTP) holds that if a subject S perceives an object O, then O is causally responsible for the perceptual experience undergone by (or had by S). That is, the object must not only have an effect on its subject, but must cause the subject’s perceptual experience.
Paul Grice reflects on our usual understanding of perception in order to derive the CTP, arguing thus: suppose some person is in the presence of a cat, and is having an experience that exactly matches the cat; but in fact, the experience was a hallucination, induced by a scientist stimulating their visual cortex. It seems reasonable to say that the person did not really “see” the cat, since the correlation between the actual circumstance and the scientist-induced hallucination was a coincidence. The best explanation for this is the absence of an appropriate causal connection between the subject and the cat. It thus follows that for a subject to see an object, the causal relationship described in the CTP must hold.
However, this argument is subject to two objections. The first, made by Smith and Jones, holds that even if some object is causally responsible for an appropriate perceptual experience in a subject, it is still possible that the subject fails to perceive the object. They take the example of sight (which is a subset of perception): a subject might look at a cat, and for some abnormal reason – perhaps trauma – this causes the subject’s brain or visual system to “blow a fuse”, triggering a hallucination; and moreover the hallucination coincidentally corresponds exactly to the actual environment. By means of this so-called “deviant causal chain”, the causal condition stated in the CTP is satisfied, even though the subject failed to “see” (i.e., perceive) the object.
A second criticism, due to Paul Snowdon, holds that the CTP assumes that the nature of perceptual experience is the same in cases of veridical perception and hallucination. This is called the “common kind assumption”, and Snowdon argues that if it is discarded in favour of the disjunctivist view that seeing and hallucinating simply involve different experiences, then Grice’s argument that the difference between seeing and hallucinating is explained by the absence of an appropriate causal connection is undermined.
To elaborate on the second criticism, it is generally accepted (and I will not contest this view here) that the nature of perceptual experience is that it consists in some relation between an object being perceived, and the subject perceiving it. Theories of perceptual experience thus differ on the nature of this relationship; one such disagreement rests on the nature of the entities that are being perceived. Naïve realists hold that there are mind-independent objects, and perception is a relation to these objects. This view entails that the common kind assumption is false: in the case of hallucination, there is no mind-independent object a perceiver is relating to, so the nature of veridical experience is qualitatively different to the nature of hallucinatory experience.
I shall deal with these criticisms in turn. The first, that the conditions of the CTP are insufficient due to the possibility of “deviant causal chains” does not refute it, but merely shows that it must hold that in addition to the object causing the appropriate perceptual experience, it must also cause it in an appropriate way. In the initial argument for the CTP given by Grice, we may well have postulated that the cat rather than the scientist stimulated the visual cortex (perhaps it walked over a remote) to give rise to the hallucination; clearly, this scenario is not qualitatively different in respect of important variables. In normal perceptual experience, the causal chain involves a direct and reliable connection between the external object and the subject’s perceptual system; in short, reinforcing the CTP by explicitly disallowing deviant causal chains is an acceptable move, since they are just as coincidental as accidentally veridical hallucination.
Turning then to the second criticism, we see that for the CTP to be true, it must either be shown that the common kind assumption holds; or show that disjunctivism is compatible with the CTP. I shall argue the former, showing that there is good reason to prefer the sense-datum theory of perceptual experience over naïve realism and similar disjunctivist theories. This view entails the common kind assumption, since in cases of both hallucination and veridical perception, one is in a perceptual relationship with these inner objects; and hence, both have the same nature, and Snowdon’s concern is a non-issue.
The sense-datum theory of perceptual experience is that the entities that a subject relates to in perception are not mind-independent objects in the external world, but mind-dependent “inner objects”. Though unintuitive, reflecting on the case of illusions shows its utility: when one perceives an object as possessing some feature, there is something – that cannot be the object itself – presented to the subject which has that quality (for example, if a subject sees a straight pen that appears to bend due to refraction, they see something that is bent). Hence, the subject sees something that is not just the object: their own inner sense-data.
Critics such as Peter Smith and O.R. Jones have argued that the inference from an appearance of some feature to the actual existence, somewhere, of that feature (the so-called phenomenal principle) is flawed. But this is a matter of nomenclature; clearly, if in the refracted pen example, I say “there exists in my mind a bent pen”, we are simply quibbling over whether its existence in my mind confers a “real” existence. But it is patently true that whenever any two people perceive a common object, they have different perceptual experiences, even in cases where neither is hallucinating or seeing an illusion. Perception necessarily happens through the lens of particular categories of the mind; one never sees an object “exactly as it is”, for this is meaningless, else everyone would usually see everything in exactly the same, “correct way”. Once we accept that perceptions are of inner sense-data, then we see that the nature of all perceptions is the same; and so if we want to distinguish between hallucinatory experience and veridical experience, Grice’s view that the difference is the absence in the former case of an appropriate causal connection emerges unscathed.
Hence, I have shown that Grice’s conceptual reflections on what it means for a subject perceive something must be bounded by the qualification that deviant causal chains must be explicitly disallowed alongside all other forms of coincidence between hallucinatory and veridical experience. I have not considered the veracity of Snowdon’s objection that the CTP presupposes the common kind assumption, instead showing that the common kind assumption is true by means of the “argument from illusion” that the perceptions are of internal sense-data, so the nature of all perceptual experience is indeed the same. Since neither objection is problematic for Grice’s argument, the CTP is true.