And while I'm thinking of it, here is a repost of a summary and discussion of The Bad Argument (5–10 and after):

I'm going to call what some people refer to as sense data, sense impressions. Moore and Russell talked about sense data. They wanted to leave open the question of whether there exists unsensed sensibilia. I suspect, however, that you and I agree that senses are had only when there is a mind to have them, and sense impression reflects this better.

I'm also going to need the distinction between intentional content and intentional objects. Two perceptual experiences can have exactly the same content, where one has an object and the other doesn't. The veridical perception of an apple has an object; the hallucination of an apple doesn't. The content of the perception is satisfied; the content of the hallucination isn't satisfied.

I actually think phenomenalism is unmotivated. For example, the main argument for it seems to be from parsimony:

(1) Phenomenalism is more economical than indirect realism.
(2) Direct realism is incorrect.
(3) If direct realism is incorrect and phenomenalism is more economical than indirect realism, we ought to prefer phenomenalism as our best theory.
(4) Hence, we ought to prefer phenomenalism as our best theory.

There is a host of arguments for (2), but they nearly all take the form of what John Searle calls The Bad Argument:

(5) In both the veridical (good) case and in the hallucination (bad) case, there is a common element—a qualitative objective visual experience going on in the visual system.
(6) Because the common element is qualitatively identical in the two cases, whatever analysis we give of one, we must give of the other.
(7) In both the veridical case and the hallucination case we are aware of something (are conscious of something, see something).
(8) But in the hallucination case it cannot be a material object; therefore, it must be a subjective mental entity. Just to have a name, call it a “sense impression”.
(9) But by step two we have to give the same analysis for both cases. So in the veridical case, as in the hallucination, we see only sense impressions.
(10) Because in both hallucinations and in veridical perceptions we see only sense impressions, then we have to conclude that we never see material objects or other ontologically objective phenomena. So Direct Realism is refuted.[1]

The problem is that (7) conflates two senses of “aware of”. The first sense is the “aware of” in (a) I am aware of a table. It describes an experience with the presence and features of the table as its conditions of satisfaction—that is, with the table as its intentional object. The second sense is the “aware of” in (b) I am aware of a painful sensation in my hand. With it, the experience and the object I'm aware of are identical. I'm experiencing pain and the only thing I'm aware of is that pain. So, in the first sense the content and object of the experience are distinct, whereas in the second sense the content and object of the experience are identical. 

The “aware of” The Bad Argument requires is the “aware of” in (a). In the veridical case, you're aware of (say) a table. There's intentional content and an intentional object. In the hallucinatory case, you aren't aware of anything. You're having a total hallucination. There's intentional content, but no intentional object. So, we aren't “aware of” something in the relevant sense in both the veridical and hallucinatory cases. So, (7) is false.

Since (7) is false, (2) remains unmotivated. Absent other reason for accepting (2), we have no reason to accept the argument from parsimony.


[1]This formulation of The Bad Argument is thanks to Searle. I've replaced his “sense data” with “sense impression” to reflect my first paragraph.