Ray Kurzweil on Memory and Cryonics

From Cryonics Nov-Dec 2012

by Michael G. Darwin

Editor’s note: As evidence is emerging that contemporary vitrification technologies are adequate to preserve identity-critical information in the brain, critics of cryonics have tried to raise the bar by postulating that the neuroanatomical basis of memory is so fragile and transient that it cannot be captured by technologies that can successfully preserve the connectome. The online exchange that gave rise to this article is 10 years old but the topic has renewed relevance again.

By Michael G. Darwin

Following the Alcor 2002 Fifth Alcor Conference on Extreme Life Extension, Eric Drexler, Robert Bradbury and Ray Kurzweil conducted an email dialogue discussing the question of what would be required to achieve successful recovery of a fully functioning human brain from cryopreservation with intact mentation and memories.

One of the most interesting and, for me, certainly one of the most memorable quotes from that discussion was this assertion by Kurzweil:

“It’s the third requirement that concerns me; the neurotransmitter concentrations, which are contained in structures that are finer yet than the interneuronal connections. These are, in my view, also critical aspects of the brain’s learning process. We see the analogue of the neurotransmitter concentrations in the simplified neural net models that I use routinely in my pattern recognition work. The learning of the net is reflected in the connection weights as well as the connection topology (some neural net methods allow for self-organization of the topology, some do not, but all provide for self-organization of the weights). Without the weights, the net has no competence.”

This quote is memorable, and surprising, because it demonstrates an inaccurate understanding of the neurobiology of memory and learning. There are certainly many things we do not know about how long term (or short term) memories are encoded in the physical structure of the brain. However, there are some things we can pretty much rule out, and the ideas put forth by Kurzweil (and not immediately criticized by Drexler and Bradbury) are among them.

It thus seems pretty clear that within the cryonics scientific community some serious education needs to take place and, to that end, let’s dissect Kurzweil’s statement.

What are neurotransmitters (NTs) and how is their concentration in brain synapses determined?

Simply put, NTs are chemicals released at the synaptic junction which are responsible for not just the transmission of signals across the synapse, but for the “strength” of the signal transmitted. Thus, they serve a “weighting function” to signaling. How much NT gets made or released is NOT a function of the NT itself, anymore than how much smoke gets released from a fire being used to send smoke signals is a function of the smoke. NTs are smoke, they are the signal medium, not the signal itself, or the source of signal, or the signal’s strength.

Focusing on the conservation of NT levels in synapses as the key to the preservation of memory is analogous to focusing on the smoke in a smoke signal as the durable element of the underlying data set. Neurotransmitters are the smoke, the real question is, what causes the NTs to be made and released in predictable amounts and ways over long periods of time. In other words, who is controlling the amount and pattern of smoke release in a smoke signaling operation?


Simplified schematic of the expression of LTP: An increase in calcium within the dendritic spine binds to calmodulin (CaM) to activate CaM Kinase II, which undergoes autophosphorylation, thus maintaining its activity after calcium returns to basal levels. CaMKII phosphorylates AMPA receptors (AMPARs) already present in the synaptic plasma membrane, thus increasing their single-channel conductance. CaMKII is also postulated to influence the sub-synaptic localization of AMPA receptors, such that more AMPA receptors are delivered to the synaptic plasma membrane. The localization of these “reserve” AMPA receptors is unclear, and thus they are shown in three different possible locations. Before the triggering of LTP, some synapses may be functionally silent in that they contain no AMPA receptors in the synaptic plasma membrane. Nevertheless, the same expression mechanisms would apply.

The current consensus in the field of the neurobiology of learning and memory is that there is extensive biochemical change in the synapse itself, probably beginning with a process called Long Term Potentiation (LTP).

Beyond that, many questions abound. It seems clear that in addition to biochemical changes in the synapse, there are also changes in the number and in the physical type and configuration of the synapses that occur during learning and memory encoding.

Most synapses cover a small area and have a compact, roughly convex shape, such as numbers 51, 59, and 81, above. These are referred to as macular synapses. Larger synapses often exhibit ‘holes’ in the middle. These holes are regions of cell membrane devoid of the specializations characteristic of the synapse, e.g. postsynaptic density, synaptic cleft, presynaptic active zone. Synapses with holes, such as numbers 45, 46, 86, 90, 94, 96, and 100, are referred to as perforated synapses. Of the 161 synapses so far classified in the neuropil, 148 are macular, while the remaining 13 are perforated. The difference between macular and perforated synapses can be seen in electron micrographs in which the postsynaptic densities have been stained.

There are well over 140 different physical TYPES OF SYNAPSE and the myriad new connections that form during learning may use many different synaptic morphologies. What’s more, sometimes many of the synapses that initially form during encoding of learning, especially multiple synapses on the same dendrite, are pared down or disappear during what is believed to be the consolidation phase of memory encoding.

Reconstruction of ‘same-dendrite, multiple synapse boutons’ (sdMSBs) and related structures in a hippocampal brain slice. (a) The sdMSB makes a synapse with the head of one spine (x) on this section. Three of the axons (4,6,7) are visible between the spine head and the dendrite (Dend). (b) Three-dimensional reconstruction of the dendrite (gray), the sdMSB axon, and all seven axons (1–7) passing through the gap between the spines (x,y). Four of the axons (2,4,5,6) are cross-sectioned to avoid obscuring the other axons. Scale bar, 0.75 μm. [Fiala JC, Allwardt B, Harris KM. Dendritic spines do not split during hippocampal LTP or maturation. Nat Neurosci. 2002 Apr;5(4):297-8. PubMed PMID: 11896399.]

Now the really interesting thing is that synapses are not transient fluctuations in the level of a biochemical, or signaling molecule – they are complex structures made of protein and protein gets made (and maintained) only as result of signal transduction between the cell nucleus and the ribosomes: DNA > RNA > ribosomes > protein. Indeed, even the synaptic vesicles and the neurotransmitters inside them, are manufactured in the cell bodies and subsequently transported to the synapses (the “right” synapses) – all of which is presumably under nuclear control. Since memories persist at least a century in humans, it is clear that the biological structure(s) that encodes them is durable and well maintained. Put another way, the mechanism that controls and determines the pattern of smoke signaling is both robust and durable.

That’s where things get hazy about how long term memory is preserved.

Our current understanding of the gating mechanisms of synapse firing suggests that the character, quantity and configuration of synapses is how memory “works,” or is encoded. But what we don’t know is how the instructions to maintain those synaptic configurations are initially activated and ultimately encoded in the neuron itself.

At this point, it should hopefully be clear that in theory, it should be possible to recover a brain with memories and personality intact, even if there was not a single molecule of NT present in any synapse, anywhere. The NTs are MADE by the neurons and released by the synapses in the “right” amount at the “right” time and in the “right” way as a function of the UNDERLYING synapse and nerve cell structure. Not the other way around!

Since I spent a good part of my adult life wrestling with the problem of ischemia-reperfusion injury in the mammalian brain, I would be remiss if I did not point out that in ischemia, NTs leak out of the boutons in the synapses. In the case of the excitatory NTs, such as glutamate, the superabundance of NTs that are present when circulation is restored causes enormous injury. The point is that NTs aren’t stable and don’t sit around in synapses without active pumping going on (requiring metabolism). They leak out rapidly under conditions of ischemia and hypoxia. And yet, if the animal or person survives (along with their brain cells), they still have intact long term memories. The neurons simply re-synthesize and replace the necessary synaptic vesicles/boutons containing the requisite types and amounts of NTs.

The question that should be preoccupying cryonicists is whether there are sufficient intact neurons and synapses present to be preserved in the first place – not whether or not the NTs levels are conserved in synapses following cryopreservation.

A scientifically sound and considerably more rigorous discussion of the issue of whether memory (and personal identity) survive cryopreservation can be found in the ChronoSphere blog article “Does Personal Identity Survive Cryopreservation?