Memory

(The forms and functions of memory in science. )

In psychology, memory is the faculty of the mind to record, preserve, and recall past experiences. His research is carried out by different disciplines: cognitive psychology, neuropsychology, and psychoanalysis.

Cognitive psychology

(Pyramid of 5 memory systems. )

The classic cognitivist current usually refers to the process of encoding, storing and retrieving mental representations. A lot of memory research in cognitive psychology involves identifying and describing its different components. To do this, psychologists rely on experimental data and on symptoms exhibited by brain-injured patients (neuropsychological data).

Modal model of memory

The most influential structural model of memory is the modal model, which divides memory into three subsystems: sensory register, short-term memory, and long-term memory. This model is a synthesis of many experimental results and represents the dominant conception of human memory in the cognitive psychology of the late 1960s. A classic formulation of this model has been proposed by Atkinson and Shiffrin (1968).

The three components of memory in the modal model are:

• Sensory register: it can hold a large amount of information in visual form for an extremely short time (a few milliseconds). This process is different from the phenomenon of visual remanence.
• Short-Term Memory (STM) (or working memory): it contains a limited number of elements, stored in verbal form for a few seconds.
• Long-Term Memory (LTM) is our intuitive design of memory. LTM does not have in practice limits of capacity or duration of memorization.

For Atkinson and Shiffrin, the probability of memorization in long-term memory (that is to say of a sustainable learning) depends solely on the duration of presence in short-term memory.

The concept of short-term memory was then deeply renewed by the concept of working memory.

Work memory (WM)

For the modal model, STM plays a special role in cognition and especially in learning new information. Experimental evidence for this function is limited, however.

Baddeley model

Given the difficulties of this model, and particularly to account for the dynamic properties of STM, Alan Baddeley and his colleagues proposed a new working memory model consisting of a central administrator and three subsystems;

• The central administrator: an attentive mechanism of control and coordination of slave systems (phonological loop and visuospatial notebook). It integrates the information from the two subsystems and puts them in touch with the knowledge stored in long-term memory. For this, it has an episodic buffer zone, which allows it to gather information, whether from sensitive impressions or long-term memory.

The three subcomponents of Baddeley and Hitch’s model are:

• Phonological loop (PL): it is able to retain and manipulate information in verbal form.
• Visuospatial notebook (VSN): it is responsible for information coded in visual form.
• Episodic buffer (EB): it allows the information contained in WM to change to LTM and vice versa. The information remains accessible to the WM, which still has to handle it.

Cowan model

Another author, Cowan (1988), has developed his own theory and model of working memory. According to Cowan, working memory is only the activated part of the LTM. Cowan, unlike Baddeley, is therefore in a unitary vision of the WM. In other words, there would be no specific structural difference, but only functional differences that would make it possible to account for the different “modules” or operations of the WM. According to this author, the most activated part of working memory corresponds to what he calls attentional focus. Indeed, the attention paid to some of the activated information would be dependent on the degree of activation of the latter, either by the perception, in the form of stimuli, or in the form of information recovered by the priming phenomena. In other words, the less information is activated, the less likely it is to be part of an explicit, verbal or pictorial representation.

The different types of memories described by Baddeley find their explanation in the amount of resources or cognitive energy that could be solicited by the entire cognitive system. Thus, this quantity of more or less limited energy would be directed towards “poles of attraction” corresponding to the most “central” zones compared to an occurrent context: lived situation, thematic, particular reasoning, domain of knowledge. The centrality of an information, or item, is measured proportionally to its familiarity (frequency of occurrence) in a domain, and by its connectivity, or the number and strength of the relationship between the item and other information from the same domain.

Cowan’s memory is, strictly speaking, a model of the connectionist and automatism type: there is only one structure composed of strongly interrelated units coupled with an energetic function, representing activation, which is localized in certain areas of the unit network as needed. This model is automatic since it does not use any control or supervisory structures either: the physical and mathematical properties of the network, the units and the energy function are sufficient to account for all the elements described by Baddeley.

Long-term memory: implicit memory and explicit memory

(Taxonomy of long-term memory systems.)

Long-term memory is the mnemic subsystem having an indefinite storage capacity in time and volume.

Psychologists divide long-term memory into two functional subsystems: explicit (or declarative) memory and implicit (or non-declarative) memory. The distinction between implicit and explicit memory concerns the use of consciousness during recall. The information in explicit memory is that which is retrieved during a conscious recall, while the implicit memory is that of behaviors and actions that do not involve consciousness.

This distinction between implicit and explicit memory partially overlaps the distinction between declarative and non-declarative memory. The declarative memory is responsible for the memorization of all the information in verbal form, that is to say those which can be expressed with the language. The notion of implicit and explicit memory generalizes this distinction to all types of information processing related to human cognition. In other words, there are automatisms for verbal, pictorial, sensory and gestural information as long as there are mental representations that can be manipulated by consciousness and attention, on which decisions can be made.

A decision refers to the conscience: to make a decision corresponds to authorize or on the contrary to inhibit a preexisting automatic process. In contrast to current assumptions, decision-making does not “create” new information itself, nor does it allow it to be retrieved: it simply allows for a last verification process on processes that have already been triggered and already activated and pre-structured information.

As for all domains related to human cognition, two theories clash to account for the distinction between implicit and explicit: one structural and the other functional. Structural theory explains the difference implicit/explicit by a difference of physical nature: explicit and implicit correspond to the solicitation of different modules and brain structures. Functional theory assumes, on the contrary, that there is only one “whole” corresponding to the medium of memory, but also that this whole is capable of different functions and the processing of different types of information. It would therefore be in this case the specific solicitation of different contexts, functions and information that would make it possible to account for the implicit/explicit difference.

There are different forms of implicit and explicit memory.

According to Squire, implicit memory would contain the following memory subforms:

• procedural memory, which allows the acquisition and use of driving skills such as cycling or playing sports;
• perceptive memory (one of whose manifestations is priming), responsible for learning visual forms, common sounds, etc. ;
• the memory of classic conditionings;
• the memory of non-associative learning, such as habituation and awareness.

The explicit memory, meanwhile, would be divided into several sub-memories:

• semantic memory;
• episodic memory.

Explicit memory: episodic memory and semantic memory

With regard to explicit long-term memory, several distinctions have been made between episodic memory and semantic memory and between implicit (procedural) memory and explicit (declarative) memory. In addition, many researches in cognitive psychology focus on the forms of mental representations used in long-term memory.

The idea of ​​the need for a semantic memory containing general knowledge for the perception and understanding of language has been suggested by artificial intelligence research. In psychology, Endel Tulving proposed in 1972 the distinction between semantic memory and episodic memory (memory of the events of the personal life), in particular to account for the symptoms of certain brain-damaged patients presenting disorders specific to one of these two types of memory.

Tulving SPI model

Endel Tulving (1995) has proposed a structural model of memory in which he distinguishes five hierarchically organized memory systems, both in terms of phylogenetic origin and in terms of preponderance within the cognitive system. We can recall that Sherry and Schacter (1987) defined the term memory system as the “interaction between mechanisms of acquisition, retention and recovery, characterized by certain operating rules (…), 2 systems (or more) with fundamentally different rules“.

From the oldest to the newest, he considers the following systems, each of which requires the integrity of the previous systems to work:

• Procedural memory: according to this model, it constitutes the oldest and most important memory system; its integrity is necessary for the operation of the following
• System of perceptual representation (SPR): it would contain perceptive drafts of the constituent elements of semantic memory.

These first two systems are said to be anoetic since they do not imply an awareness of the “object”.

• Semantic memory refers to the set of representations of general knowledge about the world.
• Primary memory is STM or WM. This system allows the temporary maintenance and manipulation of information.

These two systems are called noetic since they imply an awareness of the objects they treat.

• Episodic memory concerns representations of events located in time and space (context). This system is said to be auto-noetic because it involves an awareness of the subject and the subject as it perceives the object.

The SPI model (for serial, parallel and independent) argues that:

• Encoding is done serially, in one system after another, item by item.
• Storage is parallel, something that can be stored in multiple systems at the same time.
• Recovery is done independently, in the concerned system.

Unitary approach

Versace, Padovan and Nevers (2002) propose a different approach to memory. This approach calls into question the conception in multiple systems of memory, as well as the notion of representation in the abstractive value (the objective of sensory systems would be to “abstract” invariants) given to it by the classical cognitivist approach.

The multiple trace design considers that each confrontation with an event leads to the creation of a mnemic trace, which corresponds strictly to the sensorimotor (and in particular emotional) activations provoked by it. It would be the accumulation of these traces that would allow, from repeated confrontations to an object for example, in a wide range of different contexts, to extract somehow a meaning, recreated at each activation. This meaning, which is not stored as such, corresponds in a way to all sensorimotor activations related to this object, depending on the degree of connection.

Memory and located and distributed cognition

While most of the models discussed so far fall within the perspective of information processing in cognitive psychology, some authors propose a radically different view of cognition as a collective process inscribed in the social and physical environment. These diverse perspectives are generally grouped under the label located and distributed cognition.

In the field of memory, we can notably mention the work of Edwin Hutchins on flying airliners and shipping. It describes, for example, how the processing (memorization, recall, use) of a parameter such as the speed of the aircraft is distributed between the two members of the crew and the tools at their disposal in the cockpit. He thus suggests that cognitive processes are not purely individual phenomena but the result of the coordinated activity of participants and their instruments.

Meta memory

To explain the production of the concept of the temporal flow, it is necessary to appeal to the concept of meta-memory, or meta-mnesis, that is to say a memory of the memory, characterized by the memory of the variations of it. Meta-mnesis allows the mind to abstract from the present and to imagine a course of time by considering the succession of memories of its states of memory or more precisely the memory of the variations of its memory. This property would also be necessary for the construction of self-awareness.

(Theoretical curve of) Forgetting

The “theoretical curve of forgetting” follows an asymptote, which means that an individual constantly loses information (exponential decay) (though he never forgets everything).

Social psychology has shown, however, that communication plays a role in reinforcing the memories that are regularly and selectively discussed, to the detriment of others (“induced forgetfulness”). The more a speaker evokes the memory of a person or event, the more – both the speaker and the listener – will remember it. A phenomenon of synchronization and convergence of the memory can thus appear between two people or within a group or social network, which has for corollary a convergence of the “induced forgetfulness”. Groups can thus and also adopt erroneous memories or distort the individual memory and shape collective memories. It has been experimentally shown that “once a network agrees with what has happened, collective memory becomes relatively resistant to competing information.” The phenomenon of convergence of memory (and therefore of forgetfulness) reinforces the cohesion of the group (of a family for example) and could have a social role, but it can create a gap between groups and pose a problem when it comes to search for the truth (during a particular legal proceeding). “Memory forms group identity, which in turn forms memory, in a potentially vicious circle (…) two groups can converge to incompatible versions of the past. These versions can be preserved for posterity in statues and history books.”

The collective and historical memory is selective (as “History of the twentieth century easily remember the two world wars, but not the Spanish flu pandemic of 1918-20 which probably killed more than one of these two wars” ). It also evolves over time, especially since events mark memory more during adolescence and in young adults (“reminiscence peaks”); when a generation grows, the events that marked its members during their youth cancel the events that marked the previous generation, updating the collective memory.