Memories make us who we are – it’s pretty much impossible to overstate their importance to us. Psychologists acknowledge the existence of several different types of memory (Fig. 1): short-term (working) and long-term, explicit (consciously recalled) and implicit (unconscious, such as motor memory), episodic (autobiographical) and semantic (for facts). Normally when we think of memory, we think of explicit episodic and semantic memories. These are the ones disrupted by neurodegenerative diseases such as Alzheimer’s Disease. For researchers trying to understand the neural changes underlying learning and memory, animal models are often used, and through necessity it’s episodic memories that they tend to study.

Where in the brain are memories stored?

Memories aren’t stored in just one part of the brain, but memory researchers do focus on a few specific regions. Chief amongst these is the hippocampus. Studies of the patient Henry Molaison (H.M.), who had his hippocampus and some nearby structures removed for epilepsy surgery in the 1950s, revealed that the hippocampus was required for forming and storing new episodic memories, but not for motor learning; H.M. could improve his performance on motor tasks over days or weeks, yet have no memory of having ever encountered or practised those tasks. However, H.M.’s memories for distant events – those formed some time before his surgery – were intact. This indicated that long-term memories, those formed several years ago, seem to at some stage migrate from the hippocampus to another region or regions; subsequent research suggests these long-term memories may be stored in parts of the overlying cerebral cortex.

Another important memory structure is the amygdala. Whereas the hippocampus seems necessary for forming and storing explicit memories, and parts of the cerebral cortex may be the long-term repository of learned information, the amygdala appears to attach emotional significance to memories. This is obviously an important job, and the permanence of strongly emotional memories suggests that interactions between the amygdala, hippocampus and cerebral cortex are important determinants of memory stability, as described in more detail here [paywall] by Dr Oliver Baumann. More than just modulating memories, however, the amygdala also seems to be where memories for fear are stored. For this reason, researchers including Professor Pankaj Sah and Dr Timothy Bredy believe that understanding fear memory formation in the amygdala may help in combating conditions such as post-traumatic stress disorder.

Memories are stored by changing the connections between neurons

A lot of research supports the idea that our memories result from changing the strength of connections between neurons. These connections, or synapses, can be made stronger or weaker depending on precisely when and how often they’ve been activated in the past. Basically, a connection tends to get stronger if neuron A consistently makes neuron B fire an action potential (spike), and the connection gets weaker if neuron A consistently fails to make neuron B fire a spike. This was in fact a theoretical prediction of Canadian psychologist Donald Hebb in the mid-20th century that has now been proven experimentally. Lasting increases and decreases in synaptic strength are called long-term potentiation (LTP) and long-term depression (LTD), respectively. There’s no doubt that these changes – collectively referred to as synaptic plasticity – are a biological memory, in that something in the system has changed and that change is now imprinted in the neural circuit. But we’re still not absolutely sure that this sort of synaptic plasticity truly represents memories as we think of them.

So, how exactly might changing synaptic strengths constitute a memory? In the brain, any stimulus is represented by a particular pattern of neuronal activity – certain neurons become active in more or less a particular sequence. If you think of your cat, or your home, or that birthday cake you had when you were 5, different ensembles or groups of neurons become active. The (unproven but logical) idea is that strengthening or weakening synapses makes particular patterns of neuronal activity more or less likely to occur (Fig. 2). Think back to when you were 5 (reactivate those neurons!). At that age, if someone said “House” to you, you might’ve pictured a drawing of a house, nice and geometrical. Fast forward to now. If someone said the same thing, you may well picture your own house (Fig. 2), meaning that the same input evokes different responses. This is because your experience and memories have changed the connections between neurons, making the old “house” ensemble less likely to occur than the new “house” ensemble. In other words, to recall a memory, we have to re-activate a particular group of neurons. The idea is that by previously altering the strengths of particular synaptic connections, synaptic plasticity makes this possible.

What sorts of changes underlie synaptic plasticity?

There are various ways to strengthen or weaken synaptic connections (Fig. 3). Changes can occur on the postsynaptic side (the dendrites of the receiving neuron), such as the insertion or removal of neurotransmitter receptors, or the growth or retraction of dendritic spines, the sites where many synapses form. Presynaptic changes can also occur, such as increasing or decreasing the amount of neurotransmitter released following an action potential. All of these changes alter the postsynaptic response to the presynaptic action potential, making the postsynaptic neuron either more or less likely to fire its own spike (Fig. 3). Thus synaptic plasticity can involve modifications to already existing synapses or, more rarely, can include the growth of new connections.

The role of newly generated neurons

The idea that memories are stored by changing already existing connections between neurons has dominated research since the time of Hebb and the experiments of the 1970s that supported his proposal. However, evidence now indicates that in adult animals, particular forms of hippocampal memory can be influenced by newly born neurons. This can be thought of as cellular plasticity rather than synaptic plasticity. Such adult born neurons were first recognised in the 1960s, although it took until the 1990s for the field as a whole to accept that neurogenesis (the creation of new neurons) in adult animals could play a substantial role in brain function. Integral to this realisation was the discovery in 1992 by Professors Perry Bartlett and Linda Richards that the adult mouse brain contains neural stem cells, cells whose fate has not been determined and which can produce an endless supply of progeny, each capable of turning into any neural cell type (Fig. 4).

The hippocampus is one of two areas in the mammalian brain in which substantial levels of newly generated neurons are found in the adult; the other is the olfactory bulb, where the brain receives information from nasal sensory neurons. Within the hippocampus, newborn neurons are actually only found in one specific region, the dentate gyrus (DG). The contribution of the dentate gyrus to memory, like any other region, isn’t fully established, but it’s thought to be important in making memories precise, so that highly similar but non-identical stimuli can be distinguished during recall.

To look at whether adult-born neurons in the DG play a role in memory and learning, researchers either enhance or prevent the birth of new neurons and perform behavioural tests on the animals. Many of these studies have indeed shown that newly born neurons contribute to memory and learning, but it should also be pointed out that some studies have been unable to find such an effect. It’s likely that these discrepancies depend on experimental details such as the method(s) used to alter the levels of newborn neurons, or the specific memory task tested. For example, the inability to see an effect of neurogenesis on memory could be because the task used didn’t require the DG, where the new neurons were born. In any case, whereas more or less all researchers believe synaptic plasticity is essential for learning and memory, the case for neurogenesis isn’t yet as strong; although an involvement of newborn neurons is probable, their contribution is likely to be more restricted or more subtle than for the ever present, multi-dimensional synaptic plasticity.


Understanding memory, and being able to manipulate and improve it, is probably one of the most desired goals in neuroscience. Although I haven’t mentioned them here (I’ll cover it later), new technologies are allowing researchers to alter, delete and even implant memories in mice. That stuff is undoubtedly pretty cool, yet it comes without a really deep understanding of how exactly memories are represented in the brain. At present, synaptic plasticity is the prime candidate for a cellular mechanism of memory, but researchers don’t yet understand how—or even if—such synaptic changes translate to the circuit or network-level alterations that probably truly underlie ‘a memory’. There is also considerable interest in the role of adult-born hippocampal neurons in learning and memory. Neurogenesis declines with advancing age, yet there are ways to enhance it – for instance with exercise or environmental enrichment/mental stimulation. A role for neurogenesis in memory would therefore suggest the possibility of limiting memory loss in age-related dementias.

QBI Laboratories working on memory: Professor Perry Bartlett, Professor Pankaj Sah, Dr Timothy Bredy, Professor Jason Mattingley, Dr Judith Reinhard, Associate Professor Tom Burne

Concepts and definitions

Hippocampus – Brain structure located in the temporal lobe. Important for storing episodic and semantic memories, as well as for navigating space. Animal studies of hippocampal memory often test ‘spatial memories’..

Cerebral Cortex – The sheet of neural tissue that forms the outside surface of the brain, distinctive in higher mammals for its wrinkly appearance. The cerebral cortex is divided into horizontal layers based on anatomical and connectivity features, and is the seat of higher cognitive functions..

Amygdala – Almond-shaped structure in the brain’s temporal lobe. Important for emotional processing.

Long-term potentiation (LTP) – A long-term increase in synaptic strength. Together with long-term depression (LTD), LTP is the leading cellular candidate for memory formation.

Synaptic plasticity – The ability of a synapse to change its strength in response to patterns of activity. A change in synaptic strength means that the presynaptic neuron has either an increased or decreased ability to make its postsynaptic partner fire an action potential.

Neurogenesis – The creation of new neurons from neural stem cells. Predominantly a developmental event, but also occurs in specific brain regions during adulthood, when it is proposed to influence memory formation.

Neural stem cell – Unspecialised cell capable of dividing indefinitely and giving rise to new neurons (in neurogenesis) or glia (gliogenesis).

Author: Alan Woodruff