How does memory work? - Chapter 8

What is the taxonomy of memory?

Memory was seen as a unitary system until the 1960s. Currently several systems of memory are distinguished. These systems can be subdivided based on time or on the type of information that is stored and the way in which its retention is tested. On the basis of time, the first system of memory that can be subdivided is sensory memory (Sperling, 1960): everything that we perceive is organised by modality, for about one second (so sound remains in echoic memory, vision remains in iconic memory). However, this sensory memory fits with an observation model rather than with a memory model. The second stage of memory is working memory (Baddeley & Hitch, 1974), also known as the short-term memory (STM) (Atkinson & Shiffrin, 1968). As long as attention is paid to the information, it remains in the working memory. As soon as attention shifts, the information is only retained if it is stored in the long-term memory (LTM). What many people do not know is that the information from the STM is already stored in the LTM after approximately thirty seconds.

Long-term memory work

Based on the type of information and the method of remembering, the LTM can be subdivided into declarative (explicit) memory and non-declarative (implicit) memory (Squire, 1992). Declarative memory consists of all the memories that can be consciously evoked and verbalised: knowing what happened or is happening. Non-declarative memory consists of all the memories that cannot be verbalised but that do affect behaviour. Declarative memory is also known as explicit memory, while non-declarative memory is sometimes called implicit memory (Glas & Schacter, 1985), although this is not necessarily about the type of memory but rather on how memory is tested. If the patient is asked explicitly to retrieve a memory, this test is explicit; if the memory is tested based on the test performance, then the test is implicit. Tulving (1972) distinguished between the episodic memory (for events) and the semantic memory (for facts) of the declarative memory. There is no specific division within the non-declarative memory. The latter type includes terms such as classical conditioning, operant conditioning, the learning of skills, and priming. All of the above-mentioned dichotomy (the STM versus the LTM; declarative versus non-declarative memory and episodic versus semantic memory) are controversial.

Working memory

Baddeley's working memory model states that the 'central executive' controls the execution and coordination of the operations within the modality-specific buffers. Initially there were two of these buffers: the visual-spatial sketchpad (which briefly holds the visual information) and the phonological loop (for the temporary storage of verbal information by means of repetition or silent speech). The episodic buffer was added later mainly as a link between the working memory model and the LTM, but also as 'overflow function' to the buffer: if too much information needs to be actively retained for the phonological loop or the visuospatial sketchpad, the episodic buffer can store the excess information. In the phonological loop, verbal information can be stored for just three seconds: it can hold fewer long than short words. The capacity of the other buffers is estimated or four 'chunks' (units of information). According to Miller, you can remember 'seven plus or minus two' with four chunks. The capacity of the working memory varies greatly: some people have a capacity of just one 'chunk'.

Declarative long-term memory

Working memory is the gateway to declarative LTM, but a disturbance in the working memory (and in the attention paid to the information) does not necessarily lead to a disturbance in the storage of the LTM. The most important factor for storage in the LTM is the depth of processing. Elaboration is therefore the best way to learn something: consciously making as many associations as possible with the information that has to be learned. If much attention is paid to the information (the working memory) and many associations are made active, the information is remembered best. Three factors determine how well the information can be remembered later: how well the information is stored during learning, the retention interval (the time between encoding and testing) and the type of test. During a retention interval other information may be learned that disrupts the retrieval of the previous information (retroactive interference). Old information can also disrupt the retrieval of new information (this is called proactive interference). A 'cue' is a piece of information that is used to search through memory. There are three ways to test the LTM after a retention interval: by free recall (no cues), by cue recall (half cues) and by recognition (whole cues: the best cue is the stimulus itself). We can recognize things in two ways: by consciously remembering how we learned something (recollection) or by using a feeling of having experienced the stimuli (familiarity). We recognize more than we can reproduce ourselves. This is in line with the interference theory of forgetting: remembering becomes increasingly difficult as time goes on, because we create new memories. In this way, memories become unreachable (they do not disappear!). An alternative theory of forgetting is the decay hypothesis: memories decay or are overwritten. The difference between these two theories is difficult to test.

The difference between semantic and episodic knowledge

The distinction between semantic and episodic knowledge is not as clear as you would expect when a distinction is made between semantic memory and episodic memory. First of all, you would think that the autobiographical memory (all memories of one's own life) is completely episodic, but there is also a lot of semantic knowledge interwoven: such as the name of the place where the primary school was. Secondly, it is often a combination of semantic and episodic knowledge that ultimately determines what is remembered. Thirdly, knowledge originates from experience: all semantic knowledge comes to life from an episodic memory. The demarcation between 'semantic knowledge' and 'episodic events' can therefore be put on a continuum.

Non-declarative long-term memory

Non-declarative LTM is strongly related to classical conditioning, operant conditioning, skill learning and priming. All these forms of learning are specific and inflexible (Squire, 1992). In addition, all forms except priming are insensitive to forgetting. All these learning options are also retained in the case of an amnestic syndrome. Non-declarative memory is also known as the procedural memory and forms after a long training. The procedural memory can be tested by using a serial reaction time task (SRTT). The biggest difference between classical conditioning and operant conditioning (the latter is also called instrumental conditioning) is that with operant conditioning there is feedback by means of reinforcement or punishment. Priming is the tendency to repeat what has just been done. The most commonly used forms of priming are repetition priming and semantic priming. In repetition priming, the participant reacts faster to an actual repetition of a stimulus. In semantic priming, a person recognizes words or illustrations faster if a concept that is associated with the words is presented just before they see the actual words. 

What types of memory impairments are known?

Memory complaints

Memory complaints are the most common cognitive complaints that strongly affect the quality of life. Older people (between 55 and 85 years of age) in particular believe that their forgetfulness increases over the years (Maastricht Aging Study, 1997). The correlation between complaints and disorders is remarkably low. This can be due to (1) the high “face validity” of memory problems: they are quickly noticed in daily life, (2) the fact that many different underlying cognitive dysfunctions (such as attention problems) can underlie the memory problems, and (3) personality factors and mood problems can also play a major role in the prevalence of complaints. If a memory disorder can be objectified after neuropsychological examination, it is called an amnesia.

Amnestic syndrome

Neuropsychological patient descriptions, such as the case of HM, gave rise to a very serious memory disorder called the 'amnestic syndrome': a memory impairment that affects the learning of new information (anterograde amnesia), as well as recall of knowledge that has already been stored in memory (retrograde amnesia). There is relative preservation of the working memory and all other cognitive and intellectual possibilities (Kopelman, 2002). There seems to be a primary problem in storage, and possibly also abnormally quick forgetting (the latter, however, is only measured by reproduction tasks). Recognition is often intact, unless areas in the parahippocampal gyrus are also damaged.

Causes of memory impairments

In a patient with amnestic syndrome, one of the following two structures is often damaged: the medial temporal lobe (with the hippocampus and the parahippocampal gyrus) or the diencephalon (especially the corpora mammillaria and the anterior thalamus). An amnestic syndrome can be caused by herpes simplex encephalitis, Korsakoff's syndrome, traumatic brain injury or brain tumors where in all cases the temporal and / or diencephal structures are often damaged (Kopelman, 2002). If the disorder is caused by Korsakoff's syndrome, anterograde and retrograde amnesia are very serious. In the case of retrograde amnesia, the law of Ribot is mentioned: the most recent memories are the worst retained, while the oldest memories are best retained. The patients themselves do not seem to notice their problems due to the lack of insight into illness. Patients with the disorder due to herpes simplex encephalitis have the same disease pattern, but have more insight into the disease.

Transient global amnesia

Transient amnesia is characteristic of a post-traumatic amnesia (will be explained later) and a ‘transient global amnesia’ (TGA). More men than women suffer from TGA and often they are middle-aged. In TGA, the amnesia can last ranging from a few minutes to hours: patients are confused but the personal identity is retained. The etiology is often unclear. In a small part, epilepsy is the cause. The TGA is then called a ‘transient epileptic amnesia’ (TEA). Patients who suffer from TEA often experience gaps in their memory. Not only neurological images can lead to memory disorders; Psychiatric disorders such as depression or schizophrenia are also related to memory problems. In general, the implicit memory remains intact in amnestic patients. This is in contrast to Parkinson's or Huntington's disease (will be explained later). This can be explained by the notion that for implicit learning ability (medial) temporal structures are not used, but lower structures such as the basal ganglia are.

Psychogenic amnesia

Psychogenic amnesia (fugue) is an example of a memory disorder based on a psychological cause. Prior to the fugue there is often a persistent period of stress and / or a depressed mood, but it can also arise due to a specific situation. Patients who have had a TGA or TEA are at increased risk of developing psychogenic amnesia. The way in which stress causes psychogenic amnesia is still unclear, but it is plausible that the frontal systems do not function properly, which disrupts autobiographical knowledge. This leads to a disruption of the 'personal semantic belief system' whereby you temporarily lose your own identity (Kopelman, 2002). On the other hand, an emotional event can also lead to better remembering of a memory.

Where is memory located in the brain?

Where is working memory located?

The working memory is often related to activity in the dorsolateral prefrontal cortex, and posterior areas (Caveze & Nyberg, 2000). Through inputs from the prefrontal cortex, representations in sensory areas remain active.

Where is episodic memory located?

Almost the entire brain is actively involved in memory. For example, the frontal cortex, parts of the cerebellum and parietal lobe are active when a person searches their memory. If the memory has a specific sensory content, the sensory cortices are also active, just as the visual areas become active when retrieving a visual memory. Regions in the medial-temporal cortex, in particular the hippocampus and the parahippocampal gyrus, become very active when new memories are stored. More specifically: the hippocampus becomes active when patterns and situations are stored. The hippocampus is also very active in free reproduction. The posterior part of the parahippocampal gyrus (the parahippocampal cortex) shows a lot of activity when the space around us is stored while the anterior part of the gyrus (the perirhinal cortex) becomes active as objects are stored (this way we recognize objects). Although parts of the diencephalon do not show a clear activation pattern on memory tasks, we do know that the areas are very important: the anterior thalamic nuclei and the mammillary bodies form a functional unit with the hippocampus, and without these regions memories from the hippocampus cannot be retrieved (Aggleton & Brown, 1999).

Where is the non-declarative long-term memory located?

The non-declarative LTM also activates many parts of the brain. Conditioning is often related to activity within the basal ganglia and cerebellum. The basal ganglia are also very active when new cognitive skills are taught. In the conditioning of anxiety there is a strong involvement of the amygdala. The active brain location during priming depends on where the processing takes place (for language processing, priming takes place in the temporal lobe).