SOMETIMES, IT'S EASY: I go to an old friend's elaborate wedding in Malibu, where the seating arrangement seems explicitly designed to remind me that we no longer have anything in common. The entertainment is a 14-piece band, the closest thing to an old-fashioned "society orchestra" I've ever seen outside of a Robert Montgomery movie, equally adept at standards, Bacharach, and Israeli folk-dances. A few days later, I'm on the fourth floor of UCLA's research library, deciding which books on musical cognition are too out-of-date to bother checking out, and I realize with a start that I've been subvocalizing Henry Mancini/Johnny Mercer's "Moon River" (the lucky couple's money dance) for the last few minutes.

On other occasions, the path to an involuntary musical memory may be more idiosyncratic, but still traceable: I mention to my girlfriend that the highlight of my unremarkable day was an order of chili and spaghetti at one of the few exemplars of the once-ubiquitous Bob's Big Boy chain remaining in Los Angeles. This leads to a discussion of the combination's putative origin in Cincinnati, where it is called "two-way," "three-way," or "four-way" depending on the absence or presence of cheese and/or onions.

The subject drops, but soon enough, I'm singing, "Got kinda tired of packin' and unpackin'/Town to town up and down the dial." My mind and my mouth have somehow traveled from my lunch to the theme from "WKRP In Cincinnati" with no conscious help from me.

Most unnerving are the cases where the associational footprints, if they ever existed, are too faded or ephemeral to track: In recent days, I've been shadowed by everything from Irving Berlin's "Let's Face The Music And Dance" to Patsy Cline's "Back In Baby's Arms," to the soul oldie that goes: "Yes I'm ready (I'm ready)/To learn (to learn)/To fall in love/With you." (This is "Yes I'm Ready," by Barbara Lewis, but I had to ask a rock critic.) In all these cases, I hadn't heard the songs in question in some time, and I haven't been able to discern anything in my environment that would trigger them. It's not surprising that Berlin, arguably the 20th century's master of "the catchy tune," would be on my mind -- but why this song, and not the same composer's "You Keep Coming Back Like A Song"? And why so damn often? Are some melodies better engineered to get stuck in our heads than others, and what's going on when it happens?

EARLY PSYCHOANALYTIC SPECULATIONS about what the German language calls an Ohrwurm -- literally, "earworm" -- are about what you'd expect: remembered melodies bubble up from the subconscious, reflecting their associations with repressed fears and desires. As Theodore Reik, a student of Freud's and early exponent of psychoanalysis in America, put it in his 1953 book Haunting Melody, "Whatever secret message it carries, the incidental music accompanying our conscious thinking is never accidental." A large section of the book concerns the author's puzzlement over his own "haunting" by a so-called "Resurrection March," a choral passage from Mahler's Second Symphony at the time of the sudden death of his colleague Karl Abraham. Reik's shocking conclusion: He must have been thinking about mortality. (Does this mean that I am, in fact, ready to learn to fall in love with you?)

Unfortunately, empirical research into the causes of such occurrences is harder to come by. For reasons of experimental design, most work on music and memory necessarily focuses on subjects' ability to recognize and accurately reproduce music they've been exposed to in the laboratory. Given that individuals may have wildly disparate associations with the same piece of music, a systematic study of what "pops into" our heads seems hopeless. Still, some more general observations about how memory works can be brought to bear.

Professor Michelle Miller is a cognitive neuropsychologist, specializing in speech and language processing, who has taught at UCLA, Rice, and Northern Arizona University. (The "cognitive" just means her work is less concerned with cerebral anatomy than with "figuring out the structure of the brain by looking at how it functions.") Speaking from her office in Flagstaff, Miller told me, "There's not a huge difference between voluntary and involuntary memory, as far as we can tell. Long-term memory is best described as a vast interconnected network, but you don't get to determine all the connections. When you feel like you're voluntarily pulling a memory up, you're tracing a link back from what you're thinking about now to something else. When it happens involuntarily, you might be getting to it through a different pathway, but the same thing is happening: a feeling, a sight, or a sound drags something else up into consciousness. Voluntary control is just a small part of remembering anything."

By definition, all but our most immediate musical experiences are stored in long-term memory, which is co-ordinated by a section of the brain's temporal lobe (low and toward the back) called the hippocampus. But talking about the "location" of a particular memory, whether of a song or a person, is misleading. In his 1997 book Music, The Brain, and Ecstasy: How Music Captures Our Imagination, an accessible account of the current state of the science of musical perception, Robert Jourdain writes, "Neuroscientists have never found a place in the brain where discrete memories are stored. This is because the brain remembers things by categorizing them, not by filing away some sort of true-to-life snapshot." So even though our memories may be structured like a network, this structure doesn't seem to be realized in the brain in the straightforward way one might expect.

Additionally, Miller says, having a song running through your head, even when you're not making any sound, is more like singing than listening. "Imagining something, for the brain, is almost the same as really doing. All you subtract out are some low-level processes, in this case, the coordination of motion and breath. Your left brain might be activating representations of the words of the song and what they mean, while your right brain is calling up the tones, musical structure, and most likely your feelings and further associations -- your whole brain lights up like a Christmas tree."

Given all this, what leads our poor, song-stuffed brains to perform such acts? Just about anything, to hear Miller tell it. "Songs have such complex structures that what activates them might not be immediately apparent. It could be a past association, but it could also be related to something that's present, but not obvious. The speech rhythms of something you've overheard or read may bring up the song's rhythmic structure. Or, it could be triggered by something else you're consciously trying to remember, sort of like what we call the TOT [tip of the tongue] phenomenon, or even by environmental noise." (And once the ball is rolling, it's hard to stop: "It's probably a self-perpetuating process -- once a connection is used, it just gets stronger.") Such triggers aren't accidental in the sense of being uncaused, of course, but they're probably not what Reik had in mind. This explanation and the psychoanalytic variety aren't mutually exclusive, but it's still comforting to think that not every melody that passes through my head need be loaded with deep, symbolic freight.