Merleau-Ponty's principal treatment of hallucination comes pp. 334-345 of Phenomenology of Perception. So far as I know we yet lack a comprehensive exposition of it (cf Romdenh-Romluc 2009; Benvenuto 2015). Here at least are the main themes as I understand them:
a. The hallucinator can often tell the difference between his hallucinations and his perceptions. Merleau-Ponty gives dozens of examples of this - which basically seem to involve a bunch of early twentieth century French and German psychiatrists playing tricks on their psychotic patients with mock-ups of their hallucinations, and reporting how taken aback the patients were, and how differently they related to their real and hallucinatory experiences with the same object.
b. Hallucinations cannot be understood using either intellectualist or empiricist philosophies. Intellectualism: hallucinations are faulty judgements. Response: no, they're not. The hallucinator may or may not judge that their hallucinations are veridical experiences, but the hallucinations themselves are not judgements.
c. Empiricism: the mind relates to the body as follows. Physical stimuli affect the sense organs, give rise to neural excitation, this travels to the cortex, and then at this distal end arise sensory experiences, sense data. Hallucinations are the inner sensory experiences without their normal causes. Response: no, they're not. As in a. above, the hallucinator often readily distinguishes the two.
d. We need to understand - to 'live' - hallucination without explaining it, where by 'explaining it' he means reducing it to further items (e.g. to judgements or to sensory experiences which simply happen to obtain in the absence of their typical objects).
e. Genuine perception has built into its structure a large array of 'promises' - here Merleau-Ponty seems to draw on Husserl - that if I move over there, and my vantage changes, or if I pick this up and turn it over, that I will encounter this or that. This interconnected protentive structure of experience constitutes our normal perceptual world. Real objects also offer us what Gibson later calls sensori-motor affordances. Contrast hallucination, where there are not the same expectational pathways leading from the hallucinatory experience to other real or delusional experiences, and not the same affordances for action provided. 'I can feel swarming beneath my gaze, the countless mass of more
detailed perceptions that I anticipate, and upon which I already have a
f. So, following Minkowski: 'Hallucinations are played out on a stage different from that of the perceived world'. 'The world has lost its expressive force, and the hallucinatory system has usurped it.' They 'lack the fullness, the inner articulation which makes the real thing reside 'in itself', and act and exist by itself. The hallucinatory thing is not, unlike the real thing, packed with small perceptions which sustain it in existence. It is an implicit and inarticulate significance.'
g. 'The illusion of seeing is ... much less the presentation of an illusory object than the spread and, so to speak, running wild of a visual power which has lost any sensory counterpart. There are hallucinations because through the phenomenal body we are in constant relationship with an environment into which that body is projected, and because, when divorced from its actual environment, the body remains able to summon up, by means of its own settings, the pseudo-presence of that environment.'
h. In real experience, the perceptual landscape 'opens on to a geographical world and tends towards absolute plenitude. The normal person does not find satisfaction in subjectivity, he runs away from it, he is genuinely concerned with being in the world, and his hold on time is direct and unreflecting, whereas the suffered from hallucinations simply exploits his being in the world in order to carve a private sector for himself out of the common property world, and constantly runs up against the transcendence of time.'
i. Underlying our experience of the world is a 'momentum' or 'faith' or 'primary opinion'. This deeper functions gives perception its reality quotient. The schizophrenic lacks this. So too is this faith insufficiently established in the child who 'attributes his dreams, no less than his perceptions, to the world ... he believes that the dream is enacted ... at the foot of his bed'.
j. 'the illusory thing and the true thing do not have the same structure, for the patient to assent to the illusion, he must forget or repress the true world, and cease to refer back to it, and retain at least the ability to revert to the primitive confusion of the true and the false.'