The search for the neural correlates of memory
The search for the neural correlates of memory
The never-ending search for the engram cells
There remain other aspects to explain but that escape a materialistic paradigm with a strikingly similar pattern to that of consciousness and mentation: the neural correlates of memory. Also, in this case, one thing is certain: Memory is not stored in a specific brain area like it is on a digital computer. More than a century of research into the engram cells–that is, the group of neurons supposedly responsible for the physical representation of memory–has not led to tangible results providing convincing evidence that such cells exist. This is not a new issue. It dates back to Henri Bergson’s opposition to a reductionist understanding of memory (Bergson, 1896/1912). Bergson considered memory to be of an immaterial and spiritual nature rather than being stored in the brain.
One might assume that information content should somehow scale with brain size. This isn’t observed, however, in children who have undergone hemispherectomy (Tavares et al., 2020). How can it be that someone without half of the brain has no measurable memory impairment? Obviously, we can, as usual, explain this away by resorting to the plasticity of the brain or the functions of residual brain tissues that might save the paradigm. Or, we could conjecture that memory is stored in both hemispheres; therefore, if one hemisphere is lost, the other remains unimpaired (a hypothesis that could also fit well with supposed evolutionary advantages). Or because it is the diseased hemisphere that is removed in all these cases, Nature might have provided a mechanism that transfers the memories to the healthy hemisphere before surgery. However, we should be aware that these are conjectures, hypotheses, and speculations, not scientifically established truths. Memory storage and retrieval in biological brains remains a largely unexplained mechanism, and no conclusive evidence exists that proves it to be of a physical nature.
Other research that might suggest how and where memories are stored in brains comes from experiments performed on freshwater flatworms called planaria. These creatures can be trained to associate an electric shock with a flash of light. Therefore, one might expect that they must have encoded the experience in their brains.
Flatworm planarians have an incredible self-regeneration ability (Ivankovic et al., 2019). If this worm is cut in half, each amputated body part regenerates as two new fully formed flatworms. Not only does the part with the head form a new tail but the remaining tail also forms a new head with a brain and eyes. In 1959, James V. McConnell showed that the newly-formed planaria with a new brain also maintained its conditioned behavior (McConnell, 1959). The newly-formed living being never received the electric shock and light flash of the training phase and yet it reacted as if it had a memory of the training it had never received.
Memories, if physical, may be stored not only in the brain but also throughout the body, in non-neuronal tissue.
McConnel’s idea was that RNA molecules could transfer memory from one planarian to another as a “memory molecule.” Motivated by this idea, he injected worms with RNA taken from those trained and reported that the training had been transferred. However, further research could not convincingly reproduce McConnel’s experiments.
In 2013, Tal Shomrat and Michael Levin vindicated McConnel’s first experiments by using computerized training of planarians, replacing manual procedures that caused previous test attempts to fail (Shomrat & Levin, 2013). Then, in 2018, A. Bédécarrats had resurrected McConnel’s idea, showing how the extracted RNA from a long-term trained sea slug, the aplysia, can induce sensitization in an untrained aplysia (Bédécarrats et al., 2018). This is taken as evidence for the existence of engrams and the hypothesis that RNA-induced epigenetic changes lead to the protein synthesis required to consolidate or inhibit memory. These local translations into synaptic proteins determining the neural structure of memory are actually the mainstream engram model.
However, the problem with this hypothesis is that the fastest protein synthesis causes cellular changes in timescales of minutes. How could it possibly be responsible for our ability to store and recall memories almost instantaneously?
Moreover, the still common idea that long-time memory is mapped as synaptic connectivity is challenged by the fact that it is possible to erase synaptic connections while maintaining the same conditioned behavior in the aplysia. Long-term memory and synaptic changes can, at least in some cases, be dissociated (Chen et al., 2014). It has also been shown that the brain tissue turns over at a rate of 3–4% per day, which implies a complete renewal of the brain tissue proteins within 4–5 weeks (Smeets & al., 2018). If the synaptic trace theory is correct, and since synapses are made of proteins, how can, in the presence of this turnover, long-time memory consolidation be achieved in synaptic strengths and neural connection patterns? Notice how the fact that proteins have short lifetimes is in line with the volatility of synaptic connections. How can considerably volatile changes in synaptic connections underlie the storage of information for long periods (even in the absence of learning) (Mongillo, 2017), (Trettenbrein, 2016)? If memory is physical, other physical repositories must be shown to be viable (DNA, cellular organelles, etc.), or a paradigm shift is necessary.
The search for engrams resorts mostly to the correlation between the memory evaluation based on fear conditioning behavioral tasks of rodents and its presumed associated neural changes. For example in a series of articles the group of Tonegawa claims to have discovered engram cells (Liu et al., 2012), (Redondo et al., 2014), (Ryan et al., 2015), (Roy et al., 2016). They show how light-induced optogenetic reactivation of mice hippocampal neurons that were previously tagged during fear conditioning, induces a freezing behavior characteristic of fear memory recall. While the same activation of cells in non-fear-conditioned mice, or fear-conditioned mice in another context, did not elicit the same freezing behavior. Therefore, the activation of these context-specific neurons seems to suggest that they act like memory engrams of the specific fearful experience.
However, unclear is what really motivates the freezing behavior. The question is whether the activation led to the memory retrieval of the fearful experience leading to the freezing behavior, or if it activates first the fear-like emotional state before any memory retrieval, or if the mice might stop simply because they perceive an unexpected stimulus that might not be related with any fear or remembrance. Only the first case could potentially support the engram hypothesis, but lacking a first-person account, we will never know. While, on the contrary, the second case would only show that the activation of those cells triggers an emotional state that precedes the memory retrieval, and thus, the activated cells would not represent memory engrams (after all, we know that in humans also, stimulation of specific brain areas can lead to panic attacks associated with traumatic events, but these are not necessarily considered as the physical repository of the trauma memory.) While the third case questions whether mice freezing behavior is correlating with fear perception in the first place. A lack of motion could be due to many things, not just fear. Moreover, besides the hippocampus, it is possible to induce freezing by activating a variety of brain areas and projections, such as the lateral, basal and central amygdala, periaqueductal gray, motor and primary sensory cortices, prefrontal projections, and retrosplenial cortex (Denny et al., 2017). It is not clear what the freezing behavior is really about.
This, again, shows how, even among trained scientists, the correlation-causation fallacy is always lurking around the corner, especially when one isn’t aware of one’s own confirmation bias.
Last but not least, it turns out that almost all findings based on fear conditioning are vitiated by inadequate sample sizes, leading to questionable statistical correlations, which represent only weak evidence but then, contrary to the math, are presented as strong evidence for the engram cells hypothesis tested. At least a sample size of 15 rodents is needed to reach 80% statistical power, but in almost all cases the sample size is much smaller and a calculation is missing (Carneiro & al., 2018).
Thus, taking this all together, these lines of research are far from conclusive and, in my view, not very convincing.
However, it is, the search for the neural engram will continue, as what else could we look for? The physicalists can’t consider alternatives such as ‘extracorporeal information storage’ (for a review of this point, see (Forsdyke, 2015)). [1]
PS & Credit: Most of the sources of information I could find on this topic come from the amazing blog of Mark Mahin.
References
Bédécarrats, A. et al., 2018. RNA from trained Aplysia can induce an epigenetic engram for long-term sensitization in untrained Aplysia. eNeuro, 5(3).
Bergson, H., 1896/1912. Matter and Memory. New York: McMillan.
Carneiro, C. & al., e., 2018. Effect size and statistical power in the rodent fear conditioning literature – A systematic review. PLOS ONE.
Chen, S., Glanzman, D.L. & al., 2014. Reinstatement of long-term memory following erasure of its behavioral and synaptic expression in Aplysia. eLife, (3:e03896).
Denny, C.A., Lebois, E. & Ramirez, S., 2017. From Engrams to Pathologies of the Brain. Frontiers in Neural Circuits, 11.
Forsdyke, D.R., 2015. Wittgenstein’s Certainty is Uncertain: Brain Scans of Cured Hydrocephalics Challenge Cherished Assumptions. Biol Theory, 10, pp.336–42.
Ivankovic, M. et al., 2019. Model systems for regeneration: planarians. Development, 146(17).
Liu, X. et al., 2012. Optogenetic stimulation of a hippocampal engram activates fear memory recall. Nature, 484, pp.381-85.
McConnell, J.V., 1959. Worms and things. The Worm Runner's Digest.
Mongillo, G.R.S.L.Y., 2017. Intrinsic volatility of synaptic connections — a challenge to the synaptic trace theory of memory. Current Opinion in Neurobiology, 46, pp.7-13.
Redondo, R., Kim, J., Arons, A. & al., 2014. Bidirectional switch of the valence associated with a hippocampal contextual memory engram. Nature, 513, pp.426-30.
Roy, D., Arons, A., Mitchell, T. & al., 2016. Memory retrieval by activating engram cells in mouse models of early Alzheimer’s disease. Nature, 531, pp.508-12.
Ryan, T.J. et al., 2015. Engram cells retain memory under retrograde amnesia. Science, 348(6238), pp.1007-13.
Shomrat, T. & Levin, M., 2013. An automated training paradigm reveals long-term memory in planarians and its persistence through head regeneration. Jour. of Experimental Biology, 216, pp.3799-810.
Smeets, J. & al., e., 2018. Brain tissue plasticity: protein synthesis rates of the human brain. Brain, 141(4), pp.1122–29.
Tavares, T.P., Kerr, E.N. & Smith, M.L., 2020. Memory outcomes following hemispherectomy in children. Epilepsy & Behavior, 112(107360).
Trettenbrein, P.C., 2016. The Demise of the Synapse As the Locus of Memory: A Looming Paradigm Shift? Front. Syst. Neurosci., 10.
[1] As far as I’m aware, the authors of the above-mentioned articles don’t claim, believe, or in any way support metaphysical interpretations.