Echo (de la pensee!) what Splik said. Here are some additional thoughts:
In psychiatry, our primary job when encountering a new patient, a patient with an unclear psychiatric history, or even patients with well established psychiatric diagnoses made by good psychiatrist, is to have a diagnostic mindset. And my history taking with concurrent mental status examination (that begins the moment I lay eyes on the patient) guides this process.
So, if a patient previously unknown to me presents to the ED with "hearing voices" I start off my interview with general, open ended questions, looking for formal though disorder, negative symptoms, and delusional statements to consider a schizophrenia spectrum illness. As Splik mentioned many schizophrenics having hallucinatory experiences will adamantly deny having them when asked directly, but if I notice certain characteristic behaviors (hyperkinetic/darting eye movements, vocalizations, extreme distractibility, inappropriate laughing, etc) I will ask the patient about that directly, as Splik said. Through the course of the interview, after some rapport is built, if the patient has still not admitted to having AH, I will ask the patient a question akin to Splik's first question "is there anyone in the room besides me talking right now?" or something along the lines of "have you ever heard a voice of someone not actually present that other people couldn't hear... like someone is talking but there is no person". Keep in mind that on the initial interview whether or not the patient admits to hallucinatory experiences may not be necessary for having a general idea of what is going on. For example, one should keep in mind other symptoms of schizophrenia, subtle signs of mania even if the (prolixity of speech, elevation/irritability/lability, distraction, etc) even if the patient isn't floridly manic or depression (can sometimes be tricky on cross section to delineate from negative symptoms). Again, cross sectionally you are unlikely to come to a definitive diagnosis but at least you should be tracking symptoms to get an idea of what is going on.
As Splik mentioned, I try to gauge the presence of Schneiderian First Rank symptoms, of which the hallucinatory experiences are running commentary, voices arguing, and gedankenlautwerden (a hallucinatory voice speaking someone's thoughts out loud simultaneously with the thought- echo de la pensee occurs when the voice speaks the thought immediately afterwards- most psychiatrists group these together as Thought Echo, but s\Schneider's original term was Gedankenlautwerden). The other First Rank Symptoms are delusions of passivity, thought alientation (insertion, withdrawal, broadcasting), and delusional perception I will ask open ended questions "how many voices"... "do the voices talk to each other"... "what do they say" and then move to direct questions. Often times this is academic but I think important for documentation, clinical skill, and observation purposes.
Sometimes substance intoxication can be difficult to tell, but if the patient is having hallucinations from amphetamines/cocaine/whatever, he or she will often be extremely psychomotor agitated and have some manic type symptoms (or depressive if in a cocaine crash), even without dysautonomia or pupil dilation. Generally, even without a complete history, an astute psychiatrist can deliniate the subtleties. NMDA antagonist intoxication (PCP, Ketamine) is more difficult because you get the negative and thought disordered symptoms as well- they also can appear manic. This is becoming an increasing problem with marijuana (which most of our patients smoke
) as it is not infrequently contaminated with these substances.
Obviously lots of medications and medical conditions can cause hallucinatory experiences, but again, I try to keep a broad ddx in mind and try to rule out the above.
In the absence of signs of SCZ/mania/depression/intoxication/medical issues and unremarkable history, I then think about pseudopsychosis or malingering. Pseudopsychosis is kind of akin to "the sound of my thoughts" but in borderlines they can hear distinct voices when they get really stressed. Some of my co residents like to use directed questions of features classically LESS likely to characterize real psychosis (voice(s) located inside the patient's head and not in a perceptual plane, lateralization of voice, continuity of voice, increasing volume with Valsalva), which I sometimes ask, but I don't find these questions particularly useful, and they are definitely not acceptable when justifying discharging a malingerer or borderline- there are plenty of other ways to do this (Splik correct me if I am wrong). There is a classic paper called The Detection of Malingering Psychosis that is a fun read on this subject. The classic teaching is that true hallucinations occur in an external plane of perception, but plenty of true schizophrenics have hallucinatory experiences inside their heads (Kraepelin described this). Again, sometimes differentiating psychosis from pseudopsychosis can be tough, but if one keeps a diagnostic mindset in mind rather than focusing solely on the complaint of "hearing voices," he or she can usually determine what is going on. Also, sometimes I will deliberately ask a malingering ASPD close ended questions about the voices, and if I get a pan positive "yes" I can document that to build my case (need much more than that to kick out a patient though). Bear in mind that some people with a true psychotic history can fake psychosis and give a textbook description as above (ie, his/her psychosis is controlled with an LAI antipsychotic, yet he or she is homeless yet has enough wherewithal to malinger.
Again, keeping a diagnostic purpose in mind is paramount. The more clinical experiences one has and the more one has read, the better and more useful his or her interview and mental status exam will be.