Moon Memory

23 July 2019

Our Provost, Professor Mike Calford, describes where he thinks he was during the exact moment that the Moon landing took place and why he might not remember it as clearly as he thought.


Twenty years ago my Mother was visiting and asked me about work. I had just finished a series of lectures on the physiological basis of memory. A series that I had given - updated, of course - to biomedical and psychology students at three universities over 15 years or so. It turned out to be the last time I gave those lectures - I miss them. I concluded the series, as always, with a discussion on flashbulb memories.

Coined by Roger Brown and James Kulik in 1977, flashbulb memories are the strong recall of place and circumstance associated with significant events; particularly those shared by most people. The concept is about recall of the trivial events associated with these collectively significant events.

A generation will remember vividly where they were, what time of day it was, who they were with, and even what they were wearing, when they first saw the pictures of a burning and collapsing World Trade Center. Teaching to 20-year-old students in the 80s and 90s I had to update the examples every few years - the shooting of John Lennon, the Challenger explosion. Mature age students fixed on the death of President John Kennedy, as the strongest example. My mother brought up hearing of the end of World War II. One year, the students aligned on the death of Kurt Cobain - I had to admit my total ignorance of him.

Many will have equally vivid memories of their circumstances 50 years ago this month, when Neil Armstrong was the first to walk on the Moon. As a 12 year-old,  I remember clearly being allowed to go home from school for the afternoon, and sitting on a polished-board floor in front of a black & white television with two school friends, drinking lemon cordial from plastic Tupperware cups. Behind me,  sat my younger brother and sister along with my Mother who was breastfeeding my youngest sister. They sat on a beautiful wooden-framed couch that we would now call mid-century-modern (what happened to that!). I can easily recreate the whole scene and I can remember the weather of that day, and the particular clock on the wall as we awaited the big event.

It was somewhat ironic that I gave the lectures on memory. As an undergraduate I toiled through a course on learning and memory which required understanding, in detail, a great deal of precise description of various learning and memory experiments and their axiomatic interpretations. Mostly these were conditioning paradigms.  Operant conditioning and classical conditioning had been studied intensively in the 4 decades before I was a psychology student and a big black book summarized every variation. It was a unsatisfying field to study as, despite all of this effort, there was little attempt at explanation or work on underlying mechanisms.

Nevertheless, from this course, I had a framework into which I could fit the implications of the rapidly developing discoveries on the plasticity of neural synapses that were being made in the 70s and 80s. Synapses are the points of transmission between neurons and operate by a pre-synaptic release of a small molecule which, if there are sufficient of them (usually by coincident release from many synapses), trigger postsynaptic discharges, or action potentials.

Although barely mentioned in the big black book, Hebb's 1949 postulation of a synaptic network theory of memory stood out. It was a rare attempt at explanation.

Essentially, Hebb placed memory as an emergent property of the network of synaptic connections in the brain: repeating the same pattern of activation as an earlier event (or a thought) was sufficient to recall it. His clever twist was to predict a concept that neurophysiologists would encounter 20 years later around strengthening, or potentiation, of synapses. Even, if ultimately it proves not to explain our capacity for memory, Hebb's postulate - of a variable strength synapse that is reinforced in its capacity by repetitive use - had great impact and is the basis for all of the neural-network based AI that now affects our lives and the machines that we use.

Hebb also set an important condition for such potentiation to serve as a basis for memory: to avoid any, and every, event of synaptic transmission setting up a memory he required that some significant transmission of a given synapse (for example, a rapid volley of discharges or coincident activation of many other synapses) was necessary  for it to be potentiated and for its probability of being repeated to increase. Thus, certain patterns of activation within a complex interconnected network can become more likely - and when activated in part, or in whole, can be a memory of the original.  

Think of rain drops on a window. Initially each has an equal chance of falling into another and finding a path down the pane, but the residual trail left by a falling drop leaves a path that is likely to be followed by subsequent drops. Soon there are a few preferred paths for falling drops to join - and new drops repeat the same paths like a memory of the earlier drops.

Modern neurophysiologists conceive of memory in the same way. The conditions for potentiation of synaptic transmission are now easily demonstrated and investigated in real networks of neurons, and are consistent with those necessary to support a network theory of memory. Furthermore, functional imaging of the active human brain with PET or MRI, whilst not nearly capable of viewing synapses or networks of neurons, has shown the same pattern of activation in memory recall as that activated by original stimulation - consistent with Hebb's postulate.

Whilst the physiological evidence is consistent with a network theory, it is still a most remarkable concept that memories emerge from repeated patterns of activation,. Going back to the window-pane analogy, there is no observer watching the pattern of activation and interpreting it. Rather it is the activation pattern itself that is the memory - but that observer-less activation concept is the same for any brain function, including consciousness. Hard to conceive of, but think about it - if the brain devoted some subpart to observer functions, who would be observing the observer, etc.? 

Another challenging concept is the persistence of less-used memories - the recall gap can be decades. Here any analogy with the implementation of a neural network in machine learning becomes tenuous. A major difference from a computer is that the brain's neural circuits have an inherent instability, which extends to the very molecules of which they are comprised. The proteins that make up neurons and synapses have a half-life of days to weeks adding a complexity of maintenance to the problem of understanding memory. Not only does the network of a memory need to survive but the variable potentiated state of the synapses within it needs to be constantly recreated, even in the absence of reinforcement. There is excellent recent work on how that can be achieved, but we don't have a completed understanding yet.

Flashbulb memories present a particular challenge for understanding the physiological basis of memory. Why are they so vivid?;  and why are trivial ancillary events captured? In laboratory testing we know that certain chemicals can enhance the consolidation of memories.

For those wishing to boost their capacity there are, unfortunately, no secret memory enhancers to be had -  good sugar levels are as effective as any. Hence, the thinking is that in the enhanced emotional state of the main event we naturally release some of these chemicals and we capture memories in a particularly strong way by linking highly potentiated neural pathways of trivial matters to those of the main events. In addition, the importance of the main event promotes a high-frequency of recalls which, through repeated activation of the same neural pathways, further enhances the memory. Sounds good, but the problem with this conception is that flashbulb memories, despite their vividness are most often wrong.

Research conducted around, and a year-or-two after,  significant events such as 9/11 or the Challenger explosion has found that the strong memories evoked by questions such as "who informed you ...", "where were you ..." , "what were you wearing ..." are actually very poor. Where they can be quantified the answers based upon these vivid memories of the trivial have a less than chance probability. There are parallels with other highly emotional memory events, such as witnessing crimes.

The implications of strong mistaken memories can be profound - but, that is a different lecture. 

So what did my Mother make of my explanation of the physiological basis memory and of my personal most vivid memory of the moon landing? "Well, that is very interesting Son, but your breastfeeding sister wasn't born until September that year!"

Think about that next time you disagree with another's recollection of a shared event.

Professor Mike Calford

Provost, ANU