There are two phases in research on forgetting. During the first phase, lasting until the first decade of the 21st century, forgetting was largely regarded as a shortcoming of memory, as a malfunction of actual memory processes, such as memory formation or retrieval. There were some notable exceptions to this view, for example the idea prominently formulated by Freud that some forgetting results from targeted repression, or inhibition, of memory. In the laboratory, this form of forgetting has been studied with intentional forgetting tasks, or directed forgetting, where human participants are instructed to forget some of the material presented to them, while remembering others. Outside this context, however, forgetting was not understood as a true function of the brain, as the result of innate, or constitutive, processes that are part of normal brain function. Forgetting was not seen as adaptive behaviour, but quite the opposite, and even in Freud's view that saw forgetting as a function of the mind, repression eventually brought people onto his couch.

This view changed during the second phase, marked by discoveries of mechanisms in the brain dedicated to erase memories. This was observed relatively independent of each other in invertebrate and vertebrate animals. In fruit flies, it was found that certain neurons use dopamine signalling to eliminate memories, so that increasing the activity of these neurons accelerated forgetting, while blocking their activity prevented forgetting. About the same time it was found in rodents that the natural loss of memory over days can be prevented by interrupting the removal of parts of neurons required for neurons to communicate with each other (receptors). Notably, this form of removal requires a signal initiating it, and it was later found that other receptors may provide this signal. Just as in the fruit fly, increasing the activity of these receptors sped up natural forgetting, while blocking their activity preserved memory from obliteration. The conclusion from these findings is that certain forms of forgetting are not the result of failures of other memory mechanisms, but the outcome of dedicated processes, which are now collectively called active forgetting processes, to distinguish them from the view that dominated the first phase of research into forgetting.

One of the first classifications of forgetting stems from the Renaissance scholar Juan Luis Vives (1493-1540), who is regarded as the “Father of Psychology”. He identified three possible causes why memory may fail us. First, a storage failure could lead to a true loss of memory, such that the storage process might not have concluded correctly, or stored memory might have been impaired somehow. Second, memory fails us because we cannot retrieve a memory that exists. Third, something might have smeared up a memory, such that it gets partially broken up and corrupted. Currently, these are still regarded as the main causes for forgetting, although the terminology has changed as well as the understanding of how memory loss arises.

When a new memory is acquired, it is for a while in a labile state, during which it can be disrupted by various interventions. However, these interventions will have no effect on a new memory after some time has passed. Something happened during this time that made the new memory resistant to these perturbations. The assumed stabilization processes implementing these changes are called memory consolidation.

This discovery is more than 100 years old, and occurred shortly after experimental memory research was established in 1881 by the German psychologist Herrmann Ebbinghaus, who famously revealed the forgetting curve. About a decade after publication of his seminal monograph in 1889, two other German psychologists, Müller and Pilzecker, found that one reason why people forget is that their new memories did not have the chance to stabilize. Teaching their participants long lists of nonsense syllables, such as “JUQ,TAV, QOL, …”, they observed that memory for these lists was poorer when they distracted the participants after a list had been presented, but not when they delayed the distraction for some time. They concluded that some mental process consolidates newly formed memories so they can be retained in the long term.

Much of memory research in the neurosciences in the past century was devoted to better understand the memory consolidation process. These efforts lead to a well-supported model of initial memory formation and stabilization. It is generally accepted now that memories depend on changes to the synaptic infrastructure connecting neurons. As will be explained below, some forms of active forgetting reverse these alterations, resulting in memory loss.

The idea that memories result from structural changes has been proposed already by Réné Descartes. Using the analogy of a wax tablet, he argued that memories are akin to markings engraved on malleable, plastic matter. Just as imprints on a wax tablet, experiences leave their traces on the mind, and just like the traces on a wax tablet, these impressions may fade over time. It is perhaps no accident that words referring to memory resonate this concept. We are told to mark what someone said, we claim that something is stuck in our mind, we note that an event left an impression on us, and so forth. The idea that memory may result from structural changes to mind or brain is intuitive, as we have many examples of long-lasting records of human thought in the engravings left behind on rocks and stones by ancient cultures.

In 1913, Edward Thorndike suggested trace decay as a reason why we forget. His idea was formulated in behaviourist terminology, arguing that if we talk about memory, we are talking about the fact that a certain situation (or stimulus) produces a certain response. This is possible because the stimulus is connected to the response, which allows the former to trigger the latter. Forgetting, according to Thorndike, serves an adaptive function. Specifically, he proposed that when the connection between a stimulus and a response is not activated for a certain time, the strength of this connection is decreased, and, eventually, the stimulus can no longer produce the response. This version of trace decay is often summarized as the principle “if you don't use it, you loose it”, and Thorndike may have found this view intuitive from his practice as an educator, where regular repetition of learned material promotes its long-term retention. Consequently, he named his explanation for forgetting the “Law of Disuse”.

Quickly, however, Thorndike's law was rejected as a plausible theory because it does not provide an actual explanation for forgetting. In a now famous critique, McGeoch argued in 1932 that the theory does not provide a mechanism producing forgetting, it merely summons time itself as the causal agent. But time is simply the horizon of observation, not the reason why we forget. In McGeoch's own words: “In time iron, when unused, may rust, but oxidation, not time is responsible.”

Despite these arguments, in the 1990s Bjork and Bjork reformulated trace decay theory in form of a new theory of disuse. They argued that there are two probabilities that need to be considered to understand why we cannot remember something. The first relates to the storage strength, or how well an item was originally learned. The second one relates to the retrieval strength, indicating how easy it presently is to access the item. According to this account, the storage strength cannot be altered, so that no item ever committed to memory will ever be erased from it. What can change over time, and in relation to use, is retrieval strength. The less often an item is retrieved, the weaker the retrieval strength will become, so that it will become progressively more difficult to access the stored item. On the other hand, retrieval causes an increase in retrieval strength, so that regularly recalled items will become easier to retrieve.

Yet, trace decay theory never gained popularity in cognitive psychology and outcomes of empirical studies have often been interpreted as evidence against this explanation for forgetting. As we will see below, recent discoveries in the neurosciences paint a different picture.

The fundamental difference of the contemporary perspective on forgetting is the notion that dedicated forgetting processes exist that are necessary for normal brain function. Forgetting is therefore understood as a constitutive part of the brain, not a deplorable shortcoming or limitation, motivating the search for forgetting processes and efforts to understand their neurobiological mechanisms and contributions to memory.