Treating psychiatric patients is hard to talk about, in the way that the soul is hard to talk about. There is something immeasurable in the work, a coalescence of mind and spirit, that resists words. I think the resistance to discuss the psychiatrist's trade is due not so much to its complexity as it is to the practitioner's fear -- my fear -- that the beauty in it will evaporate if we examine it too closely; as if, by explaining the workings of empathy, we will explain it away. Maybe that's the reason for all the infighting between those who understand mental illnesses in terms of disordered life stories and others who understand them as abnormal brain chemistry.

I used to be more comfortable with the chemistry myself. As a medical student at Johns Hopkins, the scientist in me readily accepted the notion that hallucinations sprang from too much of a neurotransmitter called dopamine and that depression was a disorder of two more called norepinephrine and serotonin. I pictured antipsychotic drugs blocking chemical receptors in the brain and antidepressants flooding them. The success of electroconvulsive therapy reassured me that mental illness was a kind of faulty chemical wiring between neurons. It made me feel vulnerable to these illnesses because accidents of biology could happen to anyone.

All of that, of course, is true. Most mental illnesses do have something to do with disordered chemicals in the brain, and faulty wiring like that can happen to anyone. Many patients do, thankfully, get better on the right combination of medicines. But as it turns out, that is not the whole story. It may not be half the story. And only recently have I learned to listen closely enough to my patients, residents of a spartan, locked ward at a Boston state hospital, for them to tell me the rest.

The rest is about what happens to people when their needs for affection and security and understanding are unmet. It is about desperate survival strategies patients use to provide some insulation against the chaos of their lives.

I learned part of this by listening to a patient, a recent immigrant in his twenties. He was committed because he heard voices that told him to do everything from setting fire to his bed to overdosing on aspirin -- both of which he had done. He told me he felt sad and that he couldn't sleep and had no appetite.

Everything fit well with what's called depression with psychotic features, an illness of mood that can include either hallucinations or bizarre and unshakable beliefs. People with the disorder not only feel sad but also can see visions or hear voices. They can come to believe they are responsible for all the evil in the world. I spent a long time picking out an antipsychotic and an antidepressant medication that I thought might work. And in a couple of weeks, the voices were gone. The patient said he felt better. I agreed he was ready to go home.

The problem was that as soon as he left the hospital, he stopped taking those medicines. Psychiatrists call that noncompliance, and the word was all over the paperwork from the emergency room he ended up in, having overdosed, again, on aspirin. The doctors made sure he was medically stable, then sent him back to my ward.

"If you don't take your pills," I explained, "the voices will come back."

"I know," he said quietly. "No voices with medicine."

"So, we'll start the medicine, again."

He looked down. "I want to keep them," he whispered.

"Keep what?"

"The voices," He noticed my surprise and smiled. "Just the good ones."

The good voices, it turns out, generally talked to each other, not to my patient. Sometimes, he couldn't even make out what they were saying. But they still made him feel less alone in his tiny rented room, a room as far from the American dream he'd heard about as it was from his homeland. My treatment had taken those voices away, along with the "bad voices" that told him to hurt himself, and he couldn't live so utterly alone.

I might not be able to, either. That's where the sadness I felt at the time must have been coming from. And that sadness was as important a therapeutic tool as the medicine. It allowed me to see how his life story had shaped his illness and complicated its treatment. He wasn't just sick; he was sick and defeated and alone. The three things had become one; he needed help with all of them.

That doesn't sound very scientific. I was so prepared to embrace the medical model that I doubted these connections even as I recognized them. But they happen and, like it or not, their utility resists biomedical interpretation. The art of psychiatry is in marrying empathy and science in service to the patient.

I learned more about this from another patient, a 30-year-old woman with schizophrenia. She was admitted after assaulting a relative. For the first week of her hospitalization, our longest meeting lasted four minutes. That was how long it took me to begin explaining again how her drug abuse was worsening her hallucinations and making her violent. As if on cue, she would stand up, announce that she wasn't ill at all and walk out of my office.

I called several of the other hospitals where she had been treated. No one knew much about her other than which medicines she'd been given. No one had been able to sit with her long enough. But buried in 10-year-old records from her first hospitalization was one short paragraph about her evolving talent as an artist.

Our next meeting wasn't about illness at all; it was about painters she admired. It lasted half an hour. Two meetings later, we were talking about the steady hand needed for portraiture and how her medicine made her shake.

"Your illness has taken a lot from you," I said. Most anti-psychotic drugs do have physical side effects, some of which can't be easily managed.

"It's taken my whole life away, really," she answered. "The drugs just help me forget."

Try to listen to the schizophrenic drug user without listening to the painter, and you might as well be in the room alone.

After enough lessons from patients like these, I've started developing what some people call a third ear. I have begun to listen not only to the patient but to myself listening to the patient, feeling connections as an omniscient observer might. I stop frequently and think, "Why is it that I feel angry sitting with this patient?" Or anxious? Or happy?

I ask the patient questions designed to fill in my own emotional blanks. And the beautiful thing is that in doing so, I've ended up learning about both of us.

Keith Russell Ablow is a resident in psychiatry at Tufts-New England Medical Centers in Boston.