The notes of an old song slide through a gap in the wall, and you’re suddenly 17 again.

At that moment, the oh-so-familiar tune you’re listening to is experienced simultaneously through your short-term memory – of the sound occurring – and your long-term recall – of being a teenager, and being in love. Both memories coexist. Both shape a perception of the present.
But what’s the difference between a short-term and a long-term memory? And how exactly do these memories, working on different timescales, come together to form our experience of the world?
New York University scientists Thomas Carew and Nikolay Kukushkin have posited that the complex order of our memories is best understood in terms of “time windows”, collectively altering the state of a brain in a “temporal hierarchy”.
Writing in the journal Neuron, Carew, a professor in NYU’s Center for Neural Science and dean of the Faculty of Arts and Science, and Kukushkin, a post-doctoral fellow in neuroscience, explain their model in terms of how we comprehend sound.
“Much like sound is broken down by the auditory system into many discrete bins of frequencies that are perceived simultaneously, an experience as a whole is parsed by the brain into many ‘time windows’ that collectively represent the past,” say the scientists.
In the paper, Carew and Kukushkin explain that organisms as diverse as sea slugs and humans all have the capacity to “play” experiences in their mind on many different time frames, with some memories forgotten after a few milliseconds, and some lasting a lifetime. The limit of these various time frames, they argue, comes about from specific deviations from homoeostasis. Put more simply, how much the experience moves them and disrupts their everyday, internal balance.
When an organism is “disturbed” in this way, its mind opens a “time window”, which is then closed when the state returns to equilibrium, or when the intensity subsides and they no longer feel an emotional attachment, as an example. Neural memory, taken as a whole, is a vast repertoire of these interacting time windows.
(Above: A Mother Delousing her Child’s Hair, by Pieter de Hooch. Source: Rijksmuseum)
But how does that hierarchy work? If memories are logged from deviations in an organism’s homoeostasis, what makes some experiences more memorable than others? When you hear an old song, for example, why would it be that you remember the time you kissed a girl to it, rather than the time you ate a pizza to it?
“It’s not a straightforward leap from the time windows to pizza, but let me try to explain,” Kukushkin tells me.
“The pizza experience”
“I think your description is quite good: memories are logged from deviations in homoeostasis,” says Kukushkin. “But deviations from homoeostasis can cause other deviations from homoeostasis. What happens very often in the brain is that some deviations don’t do much by themselves, but when combined with other deviations in particular temporal arrangement (e.g., they coincide, or they occur regularly with a defined gap), they cause new deviations that last longer.
“A drunk guy in the street is a deviation from homeostasis and so is a crazy guy”
“A drunk guy in the street is a deviation from homoeostasis and so is a crazy guy, but only when they are together does someone call the police – another deviation from homoeostasis which lasts longer and is hierarchically higher.”
Kukushkin explains that brains have an astonishing amount of levels to deal with this multitude of homoeostasis blips, allowing them to pull temporal patterns from “lower-order deviations” and store them as “higher- and higher-order deviations, that exist simultaneously”.
“That is the hierarchy. So I would amend your original description in one way: memories are not so much ‘logged from’ the deviations – they are the deviations.
“The question then becomes – what kinds of deviations correspond to what kinds of stored patterns, and hence to what memories. When you are talking about something as complex as eating pizza or kissing a girl, you are operating on very high orders of pattern extraction. A pizza is not just a single stimulus, but everything you have ever learnt about pizzas throughout your life. The taste, the shape, the physical properties, the concept, the name, the letters the name consists of, etc etc.”
In a nutshell: pizza is not one thing. Pizza is a lot of things, from the sound of the word to the taste of cheese.
“There’s a giant network of cells that represents the pizza experience through countless deviations from homoeostasis, that accumulate and influence one another throughout the lifetime. Same is true for the old song, or kissing a girl, or anything complex retained by the brain from the past. The question of why you remember this rather than that is more specific.
“In your particular example, I would guess that kissing a girl had produced a deviation in homoeostasis distinct from eating pizza – perhaps something to do with larger amounts of dopamine, or the involvement of sex hormones, or something of that kind. That deviation (let’s say a sex hormone), coinciding with the deviations produced in the auditory system by the sound of the old song, produced a longer-term, higher-order deviation: strengthened connectivity between the old song network and the girl-kissing network.
“This state of stronger connectivity is also a time window, even though it might last a very long time. This new time window was not created when you were eating pizza to the song because there was no coincidence with the sex hormone.”
Nothing but time
Kukushkin makes it clear he’s fabricating the specifics to explain the basic principle (he’s assuming I’m not sexually aroused by pizza), but his description starts to illuminate how an array of changes in our chemistry can lead to an enormous edifice of “time windows”, all working in a near-constant flux of reactions with each other.
(Above: The mollusc Aplysia californica. Source: Wikimedia Commons)
He explains that the review was inspired by the team’s experimental work on the marine mollusc Aplysia californica, which was able to memorise repeated stimulations during a “training session”, but only if those stimulations were spaced across particular intervals. How does the mollusc’s nervous system remember how much time has passed between training sessions? Turns out that the state of a cell, or an ensemble of cells, or a network, can all retain something from the past for a certain amount of time – there’s no “memory molecule” or “memory circuit”, there’s only the mass of cells, retaining deviations in their internal balance as “time windows”, whether that happens over a few years or a few milliseconds.
“In fact, time is the only physical variable that is ‘inherited’ by the brain from the external world,” the scientists conclude. “Thus, memories must be ‘made of time,’ or, more precisely, of temporal relationships between external stimuli.
“In effect, the entire biological utility of memory relies on the existence of many dimensions of homoeostasis, some shorter-term and some longer-term. The many timescales of memory represent many timescales of past experience and must be simultaneously available to the organism to be useful.”
But does that help when the song plays through the wall and you’re 17 again? No.
Disclaimer: Some pages on this site may include an affiliate link. This does not effect our editorial in any way.