© David Sponder, L.E.P., BCBA, RDI CC, Floortime C2
Executive Director, Sponderworks Children’s Services

Forms of Memory

I happen to subscribe to the theory put forth by Joaquin Fuster,[1] that what we call “memory,” is actually the process of reactivating neural circuitry previously established through experience.  In other words, it isn’t “storage” per se, which implies that the brain stores what it knows in some sort of storage area where “memories” are located.  Therefore, memory is really a matter of re-experiencing.

A memory is basically a network of neocortical neurons and the connections that link them.  That network is formed by experience as a result of the concurrent activation of neural ensembles that represent diverse aspects of the internal and external environment and of motor action.” [2]

The parallel course of perception and action (motor planning) involves connections between neurons and the traveling and re-traveling of neural circuitry.  Experience, taken in through the senses, activates neurons specialized to process forms of the information coming in through sensory reception.  Each unique experience “recruits” neurons from the activated processes (each sensory reception ‘process’ [e.g. vision, audition, etc.] represents another assembly of neurons [“circuits” or “assemblies”] that perform specialized processing tasks).  Unimodal and multimodal experience creates new neuronal assemblies.

A famous axiom in neuropsychology is, “neurons that fire together, wire together.[3]”  This means that an experience involves the firing of neurons in an ad hoc assembly or network.  This newly linked network, if traveled enough, or if emotional centers of the brain assign it importance,[4] becomes more easily activated by any ‘node’ on the network.

So for instance, a person tries a new dessert and it is absolutely scrumptious.  The dessert has an appearance (processed by visual components of the brain, or visual “processors” [that are also assemblies or circuits of neurons that developed earlier through prior experience in infancy]), a smell (olfactory processors), a taste (gustatory processors), weight and texture (proprioception processors), etc.  And it also activates emotional processors (“Wow this is good”) where the emotional parts of brain assign value.  The person experiences the dessert as a coherent, unified perceptual whole by associative processes that combine the information coming in from the senses into a temporally coherent and singular (but multifaceted) stimulus.

The ‘mental representation’ of the dessert is a newly formed network of neuronal processor circuitry forged through experience.  The network has [potential] multiple entry points or nodes that once activated, re-activate the rest of the network.  So for instance, seeing a picture of the dessert reminds the person of the feeling he or she had about it.  Or perhaps if the smell of the dessert wafts in from another room, the sight of it, the feeling about it, perhaps how the dessert sounds when you stick a fork in it, etc. can re-experienced from the singular stimulus of the smell.[5]   The re-experiencing is a re-collection of the neuronal assembly originally wired together from the original experience.

Recalling Changes Memory

Since memory is really re-experiencing and not some static form of storage, the very act of recalling the experience changes the “memory.”  With the possible exception of “flashbulb memories,” recollection in the present automatically connects the original experience (the original neural network) to stimuli entering the senses from the present.  For this reason, recalled “memories” are always inaccurate to some degree, but perfect accuracy of memory isn’t required for typical functioning.

Therefore, what we describe as memory is really more about the means of encoding and retrieval than the “storage” or retention.  We all use the term ‘memory’ as a heuristic – a means of understanding neural assemblies wired in temporal coincidence and re-activated.  The term facilitates discussion about the subject, if not scientifically specific enough accuracy.  What is “retained” is ephemeral and dynamic.

Encoding of Memory

Encoding involves the process of changing existing neural assemblies or traveling new ones in Hebbian fashion.  In the short term, this merely represents a temporary lowering of thresholds for activation of the new circuitry.  In the case of longer term memory, the DNA of the neurons change, resulting in a change in the protein structure of the cell.  This is a process called epigenesis. [6], [7]

It has to do with how information is converted from initial stimulation and sensation [8] of the senses (registration or experience without awareness), the perception (awareness and conscious experience of the stimulus), and conception (the ability to understand the experience in cognitive terms).

Take color vision for instance.  The light waves enter through the retina and eventually activate the visual cortices.

Now let us compare two people, one who is color-blind and another who is not.  Both register the color in their retinas and optic nerves (the sensation or initial registration of the stimulus). But the color-blind person has trouble with the perception, or the mental interpretation of the stimulus, which makes it impossible for him to tell the difference between green and red (minimal contrast).  Both of these people, being otherwise normal, know that red and green are “colors” (a concept) and that things sharing the same color features are called “red” (the concept of ‘red,’ or, as they say in neuroscience “the concept of redness.”).

We can think of encoding as an input process, although there is no pure input or output process in the brain.  That is because the brain engages in active processes to attend, receive, interpret, and if applicable, wire up new networks (form “memories”) in order to process the information.  Therefore, the seemingly passive process of perceiving is passive by no means.

Memory Retrieval

Retrieval is what most people think of as “memory.”  It has to do with how neural circuits are reactivated as a result of internal or external triggers.  There are two primary categories of retrieval: “explicit” and “implicit.”

Explicit Memory

Explicit memory is often called “declarative memory” and refers to “…memories (or representations) that can be declared, can be brought to mind, exist in some time frame, are characterized by recognition or recall of past experiences, and for which we are consciously aware of the encoding or retrieval.[9]

While many do, I do not consider explicit memory and declarative memory as completely synonymous.  That is because explicit memory deals with static facts and knowledge (e.g. the square root of 9; the day Lincoln was shot; your name and birthday, etc., all of which represent unchanging data), as well as with personal (autobiographical) memories of episodes, which involve subjective and dynamic appraisal (perspective) of events.

Forms of Explicit Memory

Declarative Memory

This has to do with any form of memory that can be consciously called to mind and therefore declared.  Awareness is its defining feature, which is why many do not make a distinction between declarative and explicit forms of memory.  In the following paragraphs, I will make this distinction:

Semantic Memory

This has to do with non-subjective forms of information.  Semantic memories involve the remembering of facts, details, order or sequence, steps in a process, etc. (e.g., Lincoln was the 16th President of the U.S.; lunch follows recess; 2 + 2 = 4; the recipe for chicken soup; the address of my house; the definition of words, etc.).

Semantic memory has mainly to do with static features of the environment.  These stimuli do not change and do not involve the perception of change.

The word “semantic” implies understanding of meaning.  But we have to be careful here.  We are not talking about the subjective or personally defined meaning here.  We are talking instead about objective meaning, as in dictionary definitions, descriptions of cause and effect (“more labor means more cost”) and the like.

Not surprisingly, people with autism usually show strength in this more or less unemotional form of memory.  Since things like lists, maps, answers to factual questions, mands (demands, commands, remands and reprimands, etc. and the correct v. incorrect responses these forms involve) are static forms of information and communication, many with ASD show relative strength in these forms of reasoning.  ABA teaching focuses mainly on semantic forms of memory.

Finally, declarative memory is not the same thing as declarative language, although declarative language can be used to describe declarative memories.  As described in the link, declarative language has to do with voluntary sharing of experience (the language of open systems), as opposed to the instrumental use of language, which is the language of static (or closed) systems.

Episodic Memory

Episodic memory involves remembering the factual aspects of events and/or the personal and subjective experience of the event.  Endel Tulving is generally considered the foremost expert on the subject and the father of the idea, although now there is a fairly robust body of research on the subject not only in neuroscience, but also in the field of Artificial Intelligence).  Tulving describes two forms of autobiographical memory (memory of events that a person experiences):

Autonoetic Memory

Autonoetic of remembering involves the re-experiencing (reliving) of an event, implying reactivation of sensory experiences from the past.  This involves traveling backward mentally in time.  I can tell you not only that I experienced the event (noetic), but what I thought and felt, what was relevant or not, what other people thought or intended according to me, etc.  Autonoetic memories comprise the bulk of social experience-sharing.

Noetic Memory

Noetic remembering in contrast, consciously draws on personal knowledge, but does not “relive” the past or “travel backwards in time.”[10]  For instance, I can tell you that I ordered a pizza the other day, but my memory is just a thumbnail representation of the event.  I can tell you that I did it and that I was successful, but I can hardly relive it.

People with autism tend to rely mainly on noetic memories and have very weak autonoetic memory.  For instance, if you ask a child with ASD what he did at school, many will report what they did (“I played on the swing;” “I ate corn dogs for lunch”), with little or nothing in terms of elaboration of their subjective appraisal of the events (“My new friend Johnny was there too” “I like the swings but I’m getting sick of corn dogs”).

Memory of the Future or “Mental Time Travel”

Our ability to anticipate and plan is predicated on autobiographical memory.  In his essay “Memory of the Future,” neuroscientist David Ingvar refers the idea that our anticipation of the future is based on information from the past, which we “project” into imagined future events for the purposes of planning responses.[11]  The process is in large part visual for us, probably because we are primates.  That is why the most common term we use is “imagination” because we use internal visual imagery in our minds to anticipate future events.  Importantly, the imagery we use for this process consists of not only the scenery but the actions we plan to make – whether that be the very next step in an immediate temporal sequence or long term into the future.  Memory of the future utilizes mental representations established in the past, and consists of these existing representations as they are transported mentally into the future.

Imagine going to the grocery store tonight to pick up a carton of milk.  Guaranteed, your mental representation of this future event consists of visual imagery from some grocery store you have been to the past, or some amalgam of different grocery stores you’ve been to or have seen.  Your actions: getting a carton of milk, recruit representations that were established previously from your prior actions.

Now when you physically go to the store, the actual experience will be more or less congruent with your anticipation (or the collection of representations from your past).  You could show up and find that the store burned down, which would be incongruent with your projections.

Projection also involves a trajectory of change.  If I were to imagine my future, I would have to base my anticipations on trends of action that I have observed so far (noted through my tracking of change and its direction), intentions, and other possible influences, causes and effects and the like – all emanating from my experience in the past.

Implicit memory

Implicit memory comprises forms of memory “for which the subject does not need to recall the specific learning episode and may have no conscious awareness of, or may perform in an unconscious manner; those that are procedural; and those that involve rigid and inflexible representations and responses.” [12]

Phyletic Memory or “Phylogenetic Memory”

This is commonly referred to as the “memory of the species” and comprises perceptual and motor skills that are largely inborn.  The best example would be reflexes, which are carried out by the spinal cord and brain stem.  Phyletic memory is modifiable to a very limited extent (we can keep from blinking for a while or train ourselves to control certain reflexes).

Importantly, [neurotypical] infants are born with a bias towards faces and the ability to respond to attachment behaviors and emotion signals specific to our species.  The form these behaviors take over the course of development has much to do with environmental and cultural learning, but the capacity to learn them is innate.  One can hug and kiss a snake all she wants, but it will not result in attachment behavior in the snake.

Perceptual Memory

Perceptual memory has to do with the establishment of neural circuits from basic input to the senses (from external, interoception, and kinesthetic/haptic stimuli).  This allows re-cognition of the stimulus the next time the animal experiences it or something that shares common perceptual features.  This does not require consciousness or awareness, but is always involved in more declarative processes.

Motor Memory

Motor memory “…consists of representations of motor action in all its forms, from skeletal movement to the spoken language.  It too is acquired and evoked through the senses but, once acquired, it is largely represented in the neocortex of the frontal lobe, which comprises roughly one third of the human neocortex.  The most automatic and firmly established aspects of motor memory are represented outside the neocortex, notably in the basal ganglia and the cerebellum.”    [13]

Fuster describes a motor memory process where new skills require the activation of higher processing centers in the neocortex, and then, with repeated practice or overlearning, the same skills acquired through more assiduous processes become under the control of lower structures such as the brainstem and cerebellum.  Once this happens, subconscious processes, the behaviors tend to be executed automatically (or for a lot of behaviors, without awareness – such as when we shift our attention constantly in social dynamic interaction).  We commonly refer to this as “second nature.”

Even motor programs in primary areas seem to migrate to lower levels after acquisition and practice.  Following lesion of the somatosensory cortex does not abolish the skill.  Thus, corticocortical projections from somatosensory to motor cortex play a critical role in the learning of new skills but not in the execution of existing ones.  Presumably, this is the case because the latter have become automatized and relegated to lower structures of the motor hierarchy.  …this phenomenon was first described by Jackson[14] in 1866.  [These] observations suggest the migration of motor programs from higher cortex toward lower cortical or subcortical levels as their execution becomes established.”

“…as we make our way down from prefrontal cortex to motor cortex, neuronal networks generally appear to represent motor actions that are progressively less voluntary and more automatic (i.e., stimulus-bound), less abstract and more concrete in both space and time, less new and more firmly established in the experience of the organism or the species.” [15]

This is an extremely important concept necessary to understand autism.  Observations of people with autism reveal that they tend to monitor their own actions long after the brain would be expected to migrate the skills towards subconscious control.  This explicit attention leads to the self-absorption associated with the disorder.  Subsequently, the assiduous monitoring of one’s own actions takes attention and other mental resources away from tracking the environment and tracking change. This further implies that there is a corresponding lack of epigenetic change in the neural networks responsible for executing specific skills or behaviors.

Procedural Memory

The name implies ‘memory of or for procedures,’ (as it is in cognitive science), but such knowledge can also be put into words (knowledge).  Neuroscience describes it as knowledge or memory that is not easily put into words, such as the experience of balance or riding a bike.  That is, the muscle contractions devoted to actions are perceivable but not easily described declaratively.

Procedural memories involve the ability to re-activate action sequences (or thought sequences – see my page: The Concept of Motor Planning).  It encompasses perceptual and motor memory as it is acquired in the context of parallel perception-action cycles, and the memories are executed as perception-action cycles.

Repeated execution of a skill, idea, or action sequence (all the same thing) typically results in acquired procedural memories.  Once acquired, they tend to be relatively resistant to change.

With extensive overtraining… supplementary motor area activation disappears.  …As we make our way down from prefrontal cortex to motor cortex, neuronal networks generally appear to represent motor actions that are progressively less voluntary and more automatic (i.e., stimulus-bound), less abstract and more concrete in both space and time, less new and more firmly established in the experience of the organism or the species.”[16]

The reader may wonder why I say that people with autism have difficulty moving acquired skills down the motor process hierarchy to basal and cerebellar control.  After all, Fuster’s description of stereotyped and stimulus-bound behavior fits observations of autistic-like behavior, but that is superficial.  Unless the actions are ballistic, as in Tourette’s Syndrome, these behaviors have specific functions, usually for emotion-regulation, social interactive purposes, or environmental defense (blocking out; fear-based rigidity or circular).  Tellingly, most “self-stimulatory” behaviors stop whenever the individual focuses explicit attention on a changing stimulus.  And finally, while it is true that circular behaviors represent the ultimate in repeated practice and become neurodynamic attractor states, they still seem to absorb the person to the extent that they cannot attend to anything else – implying that too much of the neocortex is still involved.

[1]     Fuster, Joaquin, M.; 199; Memory in the Cerebral Cortex: An Empirical Approach to Neural Networks in the Human and Nonhuman Primate: MIT Press

[2]     Ibid.

[3]     This is a paraphrase of a statement in his book, “The Organization of Behavior” (1949; John Wiley & Sons), is as follows: “When an axon of cell A is near enough to excite cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A’s efficiency, as one of the cells firing B, is increased.”   The phrase “neurons that fire together, wire together,” repeated often in neuropsychological papers is a paraphrasing of Hebb’s original statement and is commonly referred to as “Hebb’s Law.”

[4]     Limbic centers of the brain (mainly the amygdalae), add their own [neurochemical] contribution to the experience to insure that the circuit is more easily fired (by chemically lowering the threshold for firing).  These centers have through evolution become sensitive to a stimulus’ survival value.

”Survival value” or emotional significance is the result of a combination of inborn, genetically inherent “values” (valences) obtained through phyletic memory, as well as cultural/environmental shaping, where emotion systems develop new valences based on prior experience with the stimulus or with stimuli sharing common features.

[5]     Humans have difficulty fully re-experiencing smells, tastes, tactile and proprioceptive sensations, but they can more easily re-experiencing what they saw, heard, or felt inside.  It is interesting to note that olfaction, gestation, and tactition are “proximal” senses in that they either require actual physical contact with the stimulus, or close range stimulation.  Vision and hearing on the other hand are distal senses in that stimuli can activate these senses from a distance.   People can re-experience interoceptive or somatosensory stimuli since emotion processors are activated by (neurally networked to) every experience.

[6]     This and some of the following links are to Wikipedia, the online, editable encyclopedia.  I have checked these definitions out, as all entries into Wikipedia come from contributions from the public (usually experts in the case of science entries).  These particular articles are quite good, at least at the time when I linked this file to the entries.  They provide useful elaboration for this article.

[7]     For an in-depth account of encoding processes, I refer you to Nobel prize winning neuroscientist Eric Kandel’s “In Search of Memory: The Emergence of a New Science of Mind” (W.W. Norton; 2007)

[8]     Reflexes illustrate registration without awareness or ‘simple sensation.’  Reptile brains exist mainly on this level.  The animal is not “aware” in the sense that there is any ability to appraise stimuli and respond flexibly.  Sure, a reptile is capable of responding differentially to food v. non-food for instance, but this process is merely the activation of a phyletically derived action pattern.  The animal has little or no ability to respond flexibly – which is a sign that evaluation is a more or less one-dimensional, automated and stereotyped stimulus-response relationship.

[9]     Webb, S.J.; 2007; in “Human Behavior, Learning, and the Developing Brain: Typical Development;” Coch, D., Fischer, K., Dawson, G., Eds.; Guilford Press.  Italics mine.

Interestingly, the chapter was funded by Cure Autism Now and the National Institute of Mental Health Studies to Advance Autism Research and Treatment.

[10]    Tulving, E. (1983). Elements of episodic memory. New York: Oxford University Press.

[11]    Ingvar DH. “Memory of the future”: An essay on the temporal organization of conscious awareness. Human Neurobiology. 1985;4:127–136


Spreng, N., Levine, B.; 2006; The temporal distribution of past and future autobiographical; Memory and Cognition;  34(8): 1644–1651.

Williams JMG, Ellis NC, Tyers C, Healy H, Rose G, Macleod AK.; 1996; The specificity of autobiographical memory and imageability of the future. Memory & Cognition: pp. 116–125.

[12]    Webb, S.J.; 2007; in “Human Behavior, Learning, and the Developing Brain: Typical Development;” Coch, D., Fischer, K., Dawson, G., Eds.; Guilford Press.  Italics mine.

[13]    Fuster, J. 1999

[14]    Jackson, J.H.; 1958; Selected Writings: New York: Basic Books

[15]    Fuster, J. 1999

[16]    Fuster, J. 1999