Image: Dan Saelinger; DOMINIQUE BAYNES (prop styling)
In Brief
- For decades neuroscientists have debated how memories are stored. That debate continues today, with competing theories—one of which suggests that single neurons hold the recollection, say, of your grandmother or of a famous movie star.
- The alternative theory asserts that each memory is distributed across many millions of neurons. A number of recent experiments during brain surgeries provide evidence that relatively small sets of neurons in specific regions are involved with the encoding of memories.
- At the same time, these small groupings of cells may represent many instances of one thing; a visual image of Grandma's face or her entire body—even a front and side view or the voice of a Hollywood star such as Jennifer Aniston.
More In This Article
Once a brilliant Russian Neurosurgeon named Akakhi Akakhievitch had a patient who wanted to forget his overbearing, impossible mother.
Eager to oblige, Akakhievitch opened up the patient's brain and, one by one, ablated several thousand neurons, each of which related to the concept of his mother. When the patient woke up from anesthesia, he had lost all notion of his mother. All memories of her, good and bad, were gone. Jubilant with his success, Akakhievitch turned his attention to the next endeavor—the search for cells linked to the memory of “grandmother.”
This article was originally published with the title Brain Cells for Grandmother.
Subscribe
Buy This Issue
Already a Digital subscriber? Sign-in Now
If your institution has site license access, enter here.
Already a Digital subscriber? Sign-in Now
If your institution has site license access, enter here.
See what we're tweeting about
29 Comments
Add CommentIMO, long term memories are stored as shared composite 'snippets' of sensory derived information, linked together to form specific memories.
Reply | Report Abuse | Link to thisThis method reduces the total virtual storage space required to store all memories, as common snippets are shared by many memories.
This may in part explain the dream experience, as seemingly unrelated existing memory snippets are recalled and modified if necessary, to best represent the revised collection of memories it now serves, as new memories are migrated from short term to long term memory during periods of inactivity ('downtime').
This may also explain why perception and memory degrade when sleep is deprived, as experiences cannot be quickly stored in short term memory since it's likely full. This likely degrades all memory related task performance.
This may explain why the accuracy of memories degrades over time, as component snippets are slightly altered over time to accommodate new memories.
I submit this conjecture based on decades of practical experience with information system designs, as an efficient and effective reference model producing demonstrated properties of human memory.
As to how memories are physically stored, the location of linked snippets may be spatially distributed, but snippet storage is likely localized to optimize recall performance, but I doubt they can be represented by a single cell. Snippets could be stored redundantly to reduce susceptibility to physical storage media damage.
Sounds like the movie Eternal Sunshine of the Spotless Mind
Reply | Report Abuse | Link to thisMemory, awareness, consciousness, thought, sight, hearing, touch, smell, etc cannot be stored anywhere in the brain. For the brain is nothing more than matter (dust to dust). Why does not the chair you sit in think or the floor on which you stand think?
Reply | Report Abuse | Link to thisMatter or condensed energy is what all things of we believe we are aware of contains the above attributes. Quantum Physics has proven that neither matter or energy are real but cannot tell the why or how of our earth bound experiences consist.
Time, mathematics, music and everything we think to be real are man made inventions. There is nothing absolute about science, global warming/cooling or anything else we experience in this existence. Remember the absolute scientific fact that if you walk to far you will simply fall off the earth! That was science's absolute and now we have more scientific absolutes such as the so-called 'god particle'. The question remains, what are the small particles that make up this new imagination of man?
Reply | Report Abuse | Link to thisIt is assumed that the brain sends out individual signals to preform functions of the body. WRONG
Like condoms gathering together in the vastness of an ocean each chip, atom, molecule, cell, air, water, matter, the dark matter of space with its planets and stars are as one and it takes all of this just to move your little finger.
My science snuggles a bit closer to reality as NASA is discovering the hugeness of our universe and CERN bringing forth an understanding of just how small things can be.
comment #s 1, 6, 9, 17, 19, 22, 28, 30, 33, 34, 38,
http://www.scientificamerican.com/article.cfm?id=living-in-a-quantum-world
You continue believe in the science that is created by the finite mental abilities of man. Neither you nor me can understand an infinite wisdom that creates all that we pond scum understand as reality.
Reply | Report Abuse | Link to thisSomeday, probably after what you think of as your death, will find that the reason we believe that our reality is nothing more than a point of reference that was designed for mankind to perceive a solid reality and a real universe that does not exist.
Quantum Physics 101.
I would be grateful if some neuroscientist or neuroscience enthusiast would correspond with me. I have developed neural networks with Genetic ALgorithm weight update algorithm,and would like to have a correspondent to discuss related concepts . I am surprised i am familiar with these very advanced concepts,as i did not expect leading researchers to have the same understanding as me, an amateur in the field.Maybe...just maybe...i have reached a good conclusion on some aspects of the field by tinkering with Genetic algorithms and neural network synaptic weight update algorithms.I am mereley looking for an intellectual friend for academic discussions,i am not interested in ventures or otherwise.
Reply | Report Abuse | Link to thisSo a single neuron stays alive for eighty years to store all the information I gathered about my grandmother? Sounds unlikely!
Reply | Report Abuse | Link to thisQuestion I have asked for years:
Reply | Report Abuse | Link to thisDo we know the medium by which memory is stored?
It strikes me it must be either chemical or electrical, or a combination of the two.
It has been a few years since I had occasion to ask a neurologist in position to know, his answer was "Good question."
Back in 2010, Trevor Harley and I published a paper (Loosemore, R.P.W. & Harley, T.A. (2010). Brains and Minds: On the Usefulness of Localization Data to Cognitive Psychology. In M. Bunzl & S.J. Hanson (Eds.), Foundational Issues in Human Brain Mapping. Cambridge, MA: MIT Press) in which we took a skeptical look at the reasoning in the Quiroga et al paper, and we decided that the paper's conclusion had to be spurious.
Reply | Report Abuse | Link to thisAs we wrote: "The first thing to note about this study is the strange fact that the experimenters found some neurons that just happened to respond to the chosen pictures. Who would have thought that when you put several hundred electrodes into the brain, and then show the brain roughly a hundred different images, that some 14% of the images would score a direct hit? If the experimenters’ conclusion about sparseness of encoding is correct, the chances of finding the particular neurons that respond just to, say, Jennifer Aniston must be very small indeed. Multiply that by 14 (since, on average, fourteen out of ninety-four pictures elicited a significant response in the screening part of the experiment), and we seem to have a problem.
[...]
"If the experimenters in this study were lucky enough to find neurons that encoded for 14% of the small sample of pictures shown to subjects, then one possibility is that large numbers of duplicate neurons encode each image. This leads to the following problem: If there are so many duplicate neurons encoding image-concepts (enough to enable hits on 14% of the ninety-four pictures), then how much room is there in the brain for the many thousands of other images and concepts to which we can give a name, or that we know, or that we might ever have to distinguish, or might come across in the future? If each neuron represents only one image-concept, and if Jennifer Aniston neurons are so common that a random probe easily finds one, then how much room can there be for other stuff? And what happens when we come across a new face, or object? Do we have a bank of idle neurons waiting to be recruited for the face of the next starlet? Or do we kidnap others that have been doing other jobs that have lapsed into obsolescence?
In our paper we give an alternative framework for interpreting the results. That framework is consistent with the data, but radically inconsistent with the "grandmother neuron" suggestions touted by Quiroga et al.
Have experiment been done to demonstrate that proves that the neuron isn't just a sort of radio receptor? The information could be stored in the magnetic field of the body (or any other wave), the neuron being just the special receptor for a set of information. That would explain why upon destruction of the neuron, the data become non-retrievable but also why
Reply | Report Abuse | Link to thisa forgotten data re-emerge under special stimulation.
Your one neuron that remembers to stop at a stop light better be working!
Reply | Report Abuse | Link to thisI have had my hippocampus, amygdala and left anterior temporal lobe removed due to a tumor and epilepsy. I am still able to form long term memories. My ability to recall past events feels no different now than it did prior to the surgery. How can this be if the medial temporal lobe is required to store memories? I'm sure there are many more cases out there like mine. Just because a man lost his ability to form these memories in one case does not mean that it is true that everyone requires their medial temporal lobe. The plasticity of our brains should have been touched upon in this article, imo.
Reply | Report Abuse | Link to thisExcept for interesting speculations, we know nothing significant about 'mind' and 'memory'.
Reply | Report Abuse | Link to thisA Cardinal Mistake
Reply | Report Abuse | Link to thisAn inability to illustrate correctly the cognitive processes may be the reason for the following paragraph:
“To understand the way a small number of cells become attached to a particular concept such as Jennifer Aniston, it helps to know something about the brain’s complex processes for capturing and storing images of the myriad of objects and people encountered in the world around us. The information taken in by the eyes first goes—via the optic nerve leaving the eyeball— to the primary visual cortex at the back of the head. Neurons there fire in response to a tiny portion of the minute details that compose an image, as if each were lighting up like a pixel in a digital image or as if they were the colored dots in a pointillist painting by Georges Seurat.”
We used to believe that eyes were something like TV cameras sending complete graphics to the visual cortex. However, such naïve picture is already refuted by many finds. The first one is the transmission capacity of optic nerve --- that can transmit only a small fraction of info gathered by retinal cells. Another one is that Nancy Kanwisher could not find a difference between imagined and actually seen images. And, as everyone can witness --- there is huge qualitative difference between imagined with closed eyes and actually seen with open.
Apart from this, the scent of freshly baked bread, for example, could trigger our memory of our grandma. A soup may taste just like the one our grandma served. And a tree shade may feel just like the one under which our grandma liked to have an afternoon nap...
Our perception is much more complex than we imagine --- especially when we add active as the attribute to it.
I like to illustrate this with ambiguous graphics. We usually see one of the alternative images in the graphics almost immediately. However, to see an alternative image in the same graphics --- we need to deliberately implement different set of perceptual actions. And, surprise surprise, this takes measurable time until we actually see the alternative image...
This delay is directly related to the delay Benjamin Libet explored. This delay has also been noticed by Descartes (Passions of the Soul) as soldiers were replacing instinctive reactions to flee into deliberate action to go into the battle...
I do congratulate authors on their work, but I do hope that they will articulate their future findings better.
Have a nice day,
Damir Ibrisimovic
A Cardinal Mistake
Reply | Report Abuse | Link to thisSince new policy on commenting automatically rejects any comment with an URL --- I need to add a clarification. (My published document can be found under title: “An Outline of the Active Perception Theory” with my name in the search criteria.)
In relation to perceptual actions, “concept cells” are active from the very beginning of the perceptual act. In combination with a lot of pre-processing within and between our retinal cells --- the cognitive process starts with activities of the most abstract “concept cells” guiding retinal pre-processing and quickly activating more specific “concept cells”...
What we are about to see is usually predicted well in advance. This explains speed of recognising stimuli in everyday life. And this also explains measurable perceptual confusion when totally unexpected stimuli hit our senses --- as we see unexpectedly a friend, for example. There is also an experiment (I cannot point to) in which stimuli were masked after exposure lasting, let us say, .03 sec. The subjects reported that they have seen an object, for example, but not that the object was an orange. Only after increasing exposure beyond .03 sec, the subjects were able to report that they saw an orange.
Have a nice day,
Damir Ibrisimovic
I would like to ask these question:
Reply | Report Abuse | Link to this1.Waves must derive from some source. What is the source of brain waves recorded by EEG?
2. Neuronal mitochondria are loaded with Fe and the cell body is so designed that bioelectricity can be conducted. Does this make a neuron a sort of "electromagnet" which can then generate waves theta, alpha etc? manjit
If you want to see the original DNA (four nucleotide) director of all life forms, you need look no longer for algorithm or mathematical solutions. For all the sciences including time and existence is a mirage.
Reply | Report Abuse | Link to thisTake a look at the following site for a thought to 80 light year long (nebula) that looks exactly like a DNA strand even with the tendrils.
The scientists believe that a magnetic force from a black hole some 800 light years away has twisted this nebula in to a DNA model but that is pure conjecture.
http://images.search.yahoo.com/search/images?_adv_prop=image&fr=ytff-tyc&va=DNA+nebula
This is probably the source of our holographic existence with the CMBR radiation source passing through this object which creates our reality.
6. jeetendrag10
Reply | Report Abuse | Link to this05:58 AM 1/18/13
"I would be grateful if some neuroscientist or neuroscience enthusiast would correspond with me. I have developed neural networks with Genetic ALgorithm weight update algorithm,and would like to have a correspondent to discuss related concepts . I am surprised i am familiar with these very advanced concepts,as i did not expect leading researchers to have the same understanding as me, an amateur in the field.Maybe...just maybe...i have reached a good conclusion on some aspects of the field by tinkering with Genetic algorithms and neural network synaptic weight update algorithms.I am mereley looking for an intellectual friend for academic discussions,i am not interested in ventures or otherwise"
Can you please give me your address and email ID. I will see what I can do.
I am personally convinced that the decoding of all cognitive/abstract thinking/sensory input storage etc will eventually be solved not by us neurochemists but by someone who understands Physics well enough and can work out the underlying mechanism:
"How neuronal bioelectric signals can be transduced into concrete messages by the cognitive centers?" I suspect that it involves the use of algorithms.
jeetendra just check these options>
Reply | Report Abuse | Link to thisMore Job Postings from the Web
Jan 14 Postdoctoral Fellowships in Computational Neuroimaging Laboratory of Neuro Imaging, Ucla Los Angeles, CA
PhD in neuroscience, biomedical engineering, neuroengineering, or a closely related ... are required. Knowledge of neuroanatomy and genetics and experience preparing... more
Jan 06 Computational/Electrical Engineer Neurovigil California
at the forefront of Neuroengineering and Computational Neuroscience? If you have ... brain recording technology and advanced computational algorithms, an accurate and... more
I'm confused about the assertions being made in this article about neurons in the hippocampus being "a brain region known to be involved in memory" and the summary description in Wikipedia:
Reply | Report Abuse | Link to this"The hippocampus is a major component of the brains of humans and other vertebrates. It belongs to the limbic system and plays important roles in the consolidation of information from short-term memory to long-term memory
and spatial navigation."
Being "involved" in memory and playing an important role "in the consolidation of information from short-term memory to long-term memory" are hardly synonymous.
That specific neurons in the hippocampus fire upon the recognition of images related to a specific, individual subject does not prove that the entirety of information related to that subject are contained within those neurons, unless those hippocampus neurons do not signal any other neurons upon 'firing'.
Moreover, repeatedly asserting that the recognition of specific images constitute a "concept" encapsulated within a specific set of neurons is entirely misconceived.
As I understand, conceptualization would be demonstrated by a discrete set of neurons that responded anytime an image of any woman, or bridge, or building, for example, were shown. Images of a specific individual woman, bridge or building do not represent concepts. This article seems to misrepresent the reported findings.
An interesting side to all of this is the complex (often confusing) relation(s) between "objective" and "subjective" knowledge. As we move more and more into higher-order concepts, whatever the mechanisms, we enter the realm of abstractions with both their power and dangers of interpretation. Our brains are especially good at such abstractions (2 apples + 2 oranges = 2 fruits; 2 apples + 2 oranges + 2 hamburgers = 6 foods, and so up up the scale). Thus how we form even basic concepts, which are akin to abstractions from particular events, remains wide open to future empirical investigations - investigations which we hope we can judge "objectively". Even if we all vote that this is the way things are we get stuck with the original problem of objectivity versus subjectivity in interpretation. Abstractions (concepts) are themselves powerful even when we abstract from individual neurons or circuits, however local or distributed they may be, to their broader meanings. The story goes on.
Reply | Report Abuse | Link to thisA thought provoking article, yet would not such concept cells (or groups of neurons) present something of a barrier to perfect recall, such as demonstrated by those with Savant Syndrome? Memories and concepts appear to be rather more complexly related than suggested here.
Reply | Report Abuse | Link to thisDoes any of your recent work explain or help to explain photographic or eidetic memory ? Could they bypass short term memory and store directly into long term ? It would be interesting to observe the hippocampus of a person with an eidetic memory. Does it light up when they are looking at images ? Can they select what the memorize ?
Reply | Report Abuse | Link to this“Life is not what one lived, but what one remembers and how one remembers it in order to recount it”
Reply | Report Abuse | Link to this- Gabriel García Márquez / Living to Tell the Tale -
What about the Savant syndrome with almost infinite memory abilities? How much is know about it and how much it had help to know how our memory work?
Reply | Report Abuse | Link to thisThe title in this preview is different from the actual article (Brain Cells for Grandmother) which is further qualified in the subtitle and even in the text of the article. The authors actually correctly argue in places within the article that information is not stored in single neurons. Indeed studies of response sparseness in single cortical neurons or distributed encoding in neural networks predate these hippocampal recordings - such is the nature of non-peer reviewed articles. Hippocampal neurons being involved in the consolidation of memory would be preferentially responding to whatever stimuli are being presented in the task that person was involved with. Finding therefore a Jennifer Aniston neuron would be a foregone conclusion. If the images were of Ferraris the researchers would soon enough find neurons in the hippocampus responding to Ferraris. The hippocampus is unfortunately not the right brain region to provide insight on the distributed nature of memory storage in neural networks. Perhaps the authors could have more usefully employed their time in describing the relevance of their finding for hippocampal function itself rather than attempt to generalize their finding to the rest of the brain.
Reply | Report Abuse | Link to thisAs cells are replaced over time, how is memory retained?
Reply | Report Abuse | Link to thisNeuroscientists have been monitoring brain activities using different imaging techniques like fMRI for years. With the ability to probe individual neurons, we may be able to find out the following information that brain scans cannot provide:
Reply | Report Abuse | Link to thisa. how does memory fade/change over time?
b. is a memory always stored in the same neurons or do the neuron somehow pass on the memory to other neuron?
c. during memory retrieval, how does neuron firing propagates?
d. how much does disabling a single neuron affect memory retrieval?
and so on.
Unfortunately, the experiments described in the article do not provide any new insight into memory formation, and the discussion seem to be based on what we have already learnt from brain imaging. And the conclusion that a single neuron stores a single concept or a single person is, in my opinion, quite a stretch. For example, if an intruder breaks into a house when the occupant is watching Jennifer Aniston on TV, then I expect some of the latter's Jennifer Aniston neurons will be fired whenever he sees the intruder again. Can we call Jennifer Aniston and the intruder the same concept? I don't think so.
Is this article available online somewhere? I am citing it in a blog entry I am about to publish online, and I gave it a standard APA citation, but it would also be nice if there was some way to provide instant access to my readers. A separate article I cited from another periodical is available online, and I was able to provide a direct link to it.
Reply | Report Abuse | Link to this