The Memory Illusion

by Dr Julia Shaw

  In a nutshell, benzos are a type of depressant. Many of us have likely heard that alcohol is also a depressant. This brings to mind someone alone in a bar crying into their drink. In reality, a depressant has little to do with sadness. A depressant simply depresses, or slows down, your bodily functions. Rather than thinking sad when we hear depressant, we should think slow. Slow, like problematic walking. Slow, like slow reaction time. Slow, like falling asleep. More specifically in the case of benzos, they slow down the operation of central nervous system. This, in turn, shuts down our ability to form new memories because it affects the biochemistry of our brains.

  What, more specifically, happens when we ingest benzos that causes amnesia? According to neuroscientist Daniel Beracochea from the University of Bordeaux in France,10 benzodiazepines are particularly known as acquisition-impairing molecules. This means that they prevent the formation of the protein synthesis necessary for the biological stamping of memory, much like the drugs the rats in our earlier mentioned research were given. Benzos are also generally considered to produce only anterograde amnesia, not retrograde amnesia. This means that they do little to events encountered just before their administration, but can severely impair memory for experiences after they have been ingested.

  For those of you who just cried ‘Show me more biology!’, your wish is my command. Looking at their effect in greater detail, it seems benzos enhance the effect of the neurotransmitter GABA (gamma-aminobutyric acid). According to a review article published in 2006 by Daniel Beracochea: ‘Specifically, sedative and anterograde amnesic effects of benzodiazepines were mainly attributed to α1-containing GABA-A receptor subtypes.’ For those of us who are not medical doctors, what he is saying is that the impairment is due to changes in the sensitivity of parts of the synapse that respond to GABA. Once again, it seems that changing what’s going on in the synapse changes our ability to ever even form memories.

  In classic research settings, such as the memory studies conducted by French scientist Pierre Vidailhet since the 1990s,11, 12 the effects of benzos are typically investigated by giving participants the drug before asking them to complete a task such as memorising word lists or geometric patterns. Since benzos do not affect short-term memory, it is often hard to tell that a person is under their influence as they seem to think and act normally; however, if they are tested after a time delay, they are unable to recall the items on the lists, sometimes even forgetting that they were given lists to recall at all. Similarly, if we are given a benzo in a hospital setting before an operation, we will likely forget any conversations we had with nurses, doctors and loved ones right before, during, and right after the surgery.

  I experienced this first-hand when I was put to sleep for a hospital procedure, which was then completed more quickly than anticipated. As expected, afterwards I was conscious, talking, and coherent, but clearly unable to form memories. My partner, who had come with me, apparently asked me the same ridiculous questions every few minutes, to see whether I realised that he was just on repeat. I did not. I answered as if for the first time, every time. I also apparently kept thinking I had just woken up, and that now I was fine. This is an interesting little side effect of not being able to remember the recent past, and some patients with severe amnesia due to injury report the same thing. My partner even gave me a notepad to write down my responses to his questions, and kept flipping the page so I could not see that I had written exactly the same thing a few minutes earlier. Of course, I cannot actually remember any of this, but he did show me the notepad as corroborating evidence.

  Clearly, then, the brain chemistry which is so crucial for our ability to form memories can be manipulated by the ingestion of certain kinds of drugs. But memories are more than just biochemistry. Memories are networks.

  Laser beams

  Moving up from the tiny elements of biochemistry, we are able to talk about memory structures that we can more readily see through the use of neuroimaging techniques such as fMRI, and which can be activated in living beings. Here we are looking at neurons themselves, and the physical connections between them.

  When you experience something various parts of your brain light up, which is to say that they are activated by a tiny electrical or biochemical charge passing through them. Those same neurons then remain responsible for their particular components of your memory for that same event. For example, for a single event you may have neurons in the visual cortex responsible for keeping information about what you saw, some in the auditory lobe keeping information about what you heard, and a couple in your somatosensory cortex keeping information about what you felt. The brain is therefore faced with the difficult task of finding connections between the neurons that you lit up when you had the experience, rather than reassigning neurons that are conveniently clustered together to store the memory. This means that in order to study complex memory, we must acknowledge and study these networks of neurons.

  Scientist Gaetan de Lavilléon at ESPCI University in Paris and his colleagues took a unique approach to studying these networks. They wanted to see if it was possible to play with the protein structures that underlie our memories and to thereby change neuronal connections in vivo. Publishing their findings in 201513 in the journal Nature Neuroscience, they described an experiment in which they generated memories in ways previously not thought possible.

  To do their experiment they opened up the skulls of mice and attached electrical wires very precisely to individual cells in the pleasure centres of the brain, and to an array of other areas. They wanted to create a link between so-called place cells, also known as grid cells, and pleasure. University College London neurobiologist John O’Keefe14 actually won a Nobel Prize in 2014 for the discovery of these place cells, which act like an internal GPS, letting us map our environment and storing only this kind of spatial information.

  O’Keefe and his colleagues left wires attached to the brains of the mice while they explored their environment, and noted exactly which cells were activated when the mice were in a particular location. Once they had identified these specific location cells, they could monitor them. When the mice later went to sleep, the experimenters then waited for the location cells to be spontaneously activated while the mice were dreaming. When they noticed that the mice were dreaming about the particular location they sent a jolt to the pleasure centre. This created an artificial memory, linking the place cells of a particular location with positive emotions.

  The success of this technique was demonstrated by the behaviour of the mice. When the mice woke up, they chose to spend more time in their perceived happy place than anywhere else – even though nothing positive had actually happened there. This was seen as indicative that a false memory had been formed by changing the physical structures in the brains of the mice.

  Similarly, Steve Ramirez, the late Xu Liu and their colleagues from MIT (Massachusetts Institute of Technology) wanted to see whether they could make artificial connections between memory fragments by shining laser beams into mouse brains. In their research published in 2013 in the journal Science,15 they claim ‘We created a false memory in mice by optogenetically manipulating memory engram-bearing cells in the hippocampus.’ Optogenetics is the field of science where light is used to control neurons which have been genetically modified to be photosensitive. This is done by attaching a light-sensitive protein called channelrhodopsin-2 to neurons when they are activated. For example, we can have a mouse remember a location and then attach the protein to those particular place cells. After that point, the cells can be turned on or off by the use of blue light. It’s a bit like attaching a flip switch to cells.

  Ramirez and colleagues found that activating a small but precise ensemble of mouse neurons in this way could lead to reactivation of a memory. They were able to erroneously pair old memories with new situations, thereby generating false memories. These false memories involved mice who had previously learned to associate fear of pain with one environment having this memory activated in another
environment. They now erroneously associated pain with an otherwise non-threatening environment, essentially the opposite of the sleeping mice being given erroneous associations of pleasure with a given environment.

  They specifically targeted cells in a part of the brain known as the hippocampus, a term I generally remember by thinking of a hippo on campus. Apparently this analogy is not quite right, though, because the word hippocampus comes from the Greek name for a large mythical seahorse; the structure is so named because it is shaped a little like a seahorse. The hippocampus lies pretty much in the middle of our brain and is responsible for our ability to navigate spaces, as well as for forming long-term memories. Note, however, that I get annoyed when people claim that memories are stored in the hippocampus, because this is a gross oversimplification – as we’ve just seen, memories are stored as networks throughout the brain.

  The role of the hippocampus is more one of a mediator. According to neuroscientist Dean Burnett,16 ‘Information is channelled to the hippocampus, the brain region crucial for the formation of new memories and one of the only places in the brain where brand new neurons are regularly generated. The hippocampus links all of the relevant information together and encodes it into a new memory by forming new synapses. It’s basically like someone knitting a terrifyingly complex tapestry in real time.’

  Using optogenetics to steer how our hippocampus makes memories brings to mind the idea of Matrix- or Total Recall-type science fiction scenarios, where entire complex memories are directly implanted into people’s brains using technology. We’re not quite there yet but science is making quick strides regarding the technology we can use in the brain. Optogenetics really only moved from science fiction to science fact in 2010.

  Then, towards the end of 2015 we found ourselves at the beginning of the sonogenetics17 revolution, too. Sonogenetics involves changing cells using only sound – ultrasound, to be specific. It is perhaps too early to say where this technology could lead, but all kinds of exciting developments, along with justified ethical concerns about use and abuse, are no doubt on the horizon. From a futurology perspective, it may be possible to one day shine a light or use ultrasound on a specific part of someone’s brain to alter specific memories and thereby give them a new personal history.

  In order to contextualise all of this, it is important to emphasise the fundamental principle of memory – association is everything. It is the association between the individual memory fragments in different parts of the brain that makes what we think of as a whole memory.

  I associate, therefore I remember

  Association has been seen as a core characteristic of the mind since the earliest philosophers started trying to understand how we tick. The so-called laws of association were based on a conceptualisation set in place by Plato, and were formally written as laws in 300 BC by Aristotle, as the principles that he thought underlay all learning – learning, of course, being a process of memory.

  According to Aristotle’s On Memory and Reminiscence, there are four laws of association. The first is the law of similarity – the experience or recall of one object will elicit the recall of things similar to that object. Second is the law of contrast – the experience or recall of one object will elicit the recall of opposite things. Third, the law of contiguity – the experience or recall of one object will elicit the recall of things that were originally experienced along with that object. Fourth, the law of frequency – the more frequently two things are experienced together, the more likely it will be that the experience or recall of one will stimulate the recall of the other. We can still see these laws reflected in many of today’s conceptualisations of memory, including those which we will explore in this chapter.

  For 2,000 years these four laws were assumed to be true, but their importance was largely trivialised. At least this was the case until they were revitalised by John Locke in the 17th century, and then by Hermann Ebbinghaus in the late 19th century. Ebbinghaus was a pioneer during his time, being one of the first people to study higher cognitive functions in an experimental manner. He came up with a new way to study the development of memories, by training and testing himself on so-called nonsense syllables. Nonsense syllables are collections of letters which have no inherent meaning, such as OOB or KOJ. Ebbinghaus’s idea was that these would be easy to memorise but would have no previous associations. In other words they were chosen because they should not skew results by having any existing meaning, as pre-existing meaning would make some of them easier to remember. While this has been challenged since, with researchers arguing that even nonsense syllables can still be assigned meanings, the effort was laudable.

  In 1885, Ebbinghaus summarised his findings and published his magnum opus Über das Gedächtnis,18 later translated into English as Memory: A Contribution to Experimental Psychology.19 His experiments, which he conducted using himself as the only participant, gave us many insights into memory formation and storage that are still accepted widely.

  The modern concept of associative activation is an elaboration on these original propositions by Aristotle and Ebbinghaus, and refers to the idea that there is increased activity in particular memories when other, conceptually similar, ideas or experiences are processed. For example, if you think about swimming, you will almost certainly automatically activate memories of the associated concepts of water, pool and bathing suit. This notion assumes that individuals develop a set of frequently used words and concepts. Each individual concept or word in the brain can be called a node. These nodes can be linked to one another to create complex ideas.

  Nodes that have similar meanings are thought to have stronger connections. So, the node ‘police officer’ is likely to be very strongly associated with the node ‘law’ and very weakly associated with the node ‘table’. Once we activate a node, the energy we send out to it can be conceptualised as radiating out from that original source and then activating other related nodes automatically. So if the node we activate first is police officer, the energy will then automatically travel out from there to other strongly associated nodes such as law.

  This kind of associative activation is something that I think about every time I travel on the Underground in London, or the Metro in Paris. The network of lines could be seen to represent the brain, with each station representing a node. Just as we can get from any station to any other, so it is possible to move through a series of associations from any node in the brain to any other. Just as some stations are only a few minutes’ direct travel from one another, while to get between others we need to go through a long and complicated series of connections and changes, so some associations between nodes are far stronger and easier than others, meaning that those associated memories come more easily to us. To extend the analogy even further, sometimes rail connections fail or break down, so we end up at a station other than the one we intended. In a similar way memory errors can be envisaged as occurring due to a physiological breakdown in the connections between nodes.

  Associative activation can contribute to false memory formation at two points in time – during encoding and during recall. During encoding, it is possible to present a number of concepts to an individual, without ever mentioning the main concept. So, a researcher could mention the concepts law, man and uniform, without ever mentioning the concept police officer. However, because this concept was activated automatically due to its association with the others, it might also be encoded by the participant along with the others – they might think that a police officer was directly mentioned by the researcher when this was not the case.

  A similar kind of mistake can be made during recall of an event; when trying to remember which concepts were engaged with earlier, an individual may remember that the concepts law and uniform were mentioned, and the sense of familiarity with the concept police officer (again, because it is automatically activated) may encourage the individual to incorporate this into their recall.

  Associative activation thus implies that false memories are the downs
ide of being able to form powerful associations. The upside is that these associations allow us to have memories in the first place, along with the ability to creatively rewire ideas to respond to our environment and come up with complex solutions to problems. This also means that if associations between memories or concepts can be strengthened or weakened, this can affect the likelihood of memory illusions and errors.

  Who invited Kevin?

  This perhaps all seems a bit abstract at this point, so let us try a more lively explanation. Everyone loves a party. The physical representation of a memory in the brain is commonly known as an engram. And, this engram needs to be able to connect with other physical representations, other engrams – memory is inherently associative in nature, so all of these physical representations need to be linked to enable us to form and access our memories. This means that every time you access a memory, you are essentially having an engram party in your brain.

  Picture the scene: Engram arrives at a party. He is pleased to see that his two best friends are already there. Engram sits down and quickly reconnects with them. He has associated with these friends often and they have an incredibly strong bond – so strong that their connection feels automatic. Engram’s besties are the concepts and ideas that are inherently and strongly linked to the specific piece of information Engram embodies. Let’s imagine Engram is the memory of your favourite park. Those ‘friendly’ engrams, the most closely related concepts, are perhaps the location of your favourite pond and what the trees look like.

  The next to arrive at the party are Engram’s other friends and colleagues. Some of them he only has a weak connection with, and some of them he finds incredibly boring, but they usually still come to his parties. These are the concepts that still have connections with the memory you are activating, but they have weaker links. Perhaps these are the memories of a bench in the park, an adjacent street to the park, and the closest café. The pieces of information may not always be useful, and the associations are not necessarily strongly reinforced, but they are mostly still activated whenever the park memory is activated.