Submissão: 10/09/2019 Aprovação:
10/09/2019 Publicação: 18/12/2019
Dossiê Filosofias da memória
How to distinguish long-term individual memory representations? A historical and critical journey
Como distinguir
representações de memória individuais de longo prazo? Uma jornada histórica e
crítica
Marina
Trakas
Postdoctoral
researcher at Conicet (Consejo Nacional de Investigaciones Científicas y
Técnicas), Buenos Aires, Argentina.
Abstract: Memory
is not a unitary phenomenon. Even among the group of long-term individual
memory representations (known in the literature as declarative memory) there
seems to be a distinction between two kinds of memory: memory of personally
experienced events (episodic memory) and memory of facts or knowledge about the
world (semantic memory). Although this distinction seems very intuitive, it is
not so clear in which characteristic or set of interrelated characteristics
lies the difference. In this article, I present the different criteria proposed
in the philosophical and scientific literature in order to account for this distinction:
(1) the vehicle of representation; (2) the grammar of the verb “to remember”;
(3) the cause of the memory; (4) the memory content; and (5) the phenomenology
of memory representations. Whereas some criteria seem more plausible than
others, I show that all of them are problematic and none of them really fulfill
their aim. I then briefly outline a different criterion, the affective
criterion, which seems a promising line of research to try to understand the
grounds of this distinction.
Keywords: Kinds
of memory; Declarative memory; Episodic memory; Semantic memory; Autobiographical
memory.
Resumo: A memória não é um fenômeno unitário. Mesmo
entre o grupo de representações de memória individuais de longo prazo
(conhecidas na literatura como memória declarativa), parece haver uma distinção
entre dois tipos de memória: memória de eventos experimentados pessoalmente
(memória episódica) e memória de fatos ou conhecimentos sobre o mundo (memória
semântica). Embora essa distinção pareça muito intuitiva, não é tão claro em
qual característica ou conjunto de características inter-relacionadas reside a
diferença. Neste artigo, apresento os diferentes critérios propostos na
literatura filosófica e científica para dar conta dessa distinção: (1) o
veículo de representação; (2) a gramática do verbo “lembrar”; (3) a causa da
memória; (4) o conteúdo da memória; e (5) a fenomenologia das representações da
memória. Embora alguns critérios pareçam mais plausíveis que outros, mostro que
todos são problemáticos e nenhum deles realmente cumpre seu objetivo. Em
seguida, descrevo brevemente um critério diferente, o critério afetivo, que
parece uma linha de pesquisa promissora para tentar entender os fundamentos
dessa distinção.
Palavras-chave: Tipos de memória; Memória declarativa; Memória
episódica; Memória semântica; Memória autobiográfica
The
reason for distinguishing different categories is—if not to identify natural
kinds—at least to make distinctions that are conceptually or empirically useful
in order to think about and analyze mental phenomena. As Boyd[1]
and Machery[2]
have pointed out, entities should be clustered together when they share a large
set of properties because of some uniform causal mechanism and when these
clusters optimize the inductive and explanatory purposes of theories that make
reference to them. Although there seems to be an essential difference between
memories of experiences and memory of facts in order to justify the distinction
in terms of different (natural) kinds, it is not so clear in which
characteristic or set of interrelated characteristics lies this difference. If
the difference lies in the content, the notions of “experience” and “fact”
would need further specification. But the distinction could also be justified
through other main characteristic, such as their vehicles of representation, their
causes, or their different phenomenology.
Even if the difference in kind is grounded in the existence of distinct
memory systems, different characteristics at the personal level must supervene
from these different neural realizations. So the
problem of identifying the property or interrelation of properties that
establish a distinction between different kinds of long-term individual memory
representations still persists.
Through
history, philosophers and cognitive scientists have proposed different criteria
to account for this intuitive distinction between memories of experiences and
memories of facts. In some cases, the criterion proposed leads to make further
distinctions between long-term individual memory representations. In this
article, my aim is thus to present the different criteria that have been
proposed in the literature to distinguish different kinds of long-term
individual memory representations, and evaluate if they successfully fulfill
their aim. I distinguish five types of criteria according to (1) the vehicle of
representation; (2) the grammar of the verb “to remember”; (3) the cause of the
memory; (4) the memory content; and (5) the phenomenology of memory
representations. As this analysis shows, whereas some criteria are more plausible
than others from a theoretical and experimental perspective, none of them is
exempt of problems. The major inconvenient seems to be focused on a poor
analysis of the interaction and correlation between different significant
properties of our memory experiences. At the end of this article, in section
(6), I briefly outline a different criterion, the affective criterion, which is
promising for accounting for this correlation and may thus in the future, with
further research, prove to be more successful than the others to elucidate the
diversity of our long-term individual memory representations.
One
important remark: as I have already suggested, this analysis is confined to
long-term individual memory representations and thus, excludes criteria that
contrast these memory representations with something else, such as short-term
representations, procedures, habits and motor skills, or collective memories.
Therefore, the aim of this work is to set out and critically analyze the
variety of criteria present in the literature to distinguish different kinds of
long-term individual memory representations.
The grammatical criterion
The grammatical criterion—a criterion based on
the grammar of the verb “to remember”—is in certain way the most superficial
criterion to distinguish different long-term individual memory representations.
The distinction between different kinds of memories according to the
grammatical objects of the verb to remember has its origins in the distinction
between memories of facts and memories of experiences that was object of
philosophical debate during the 60’s and 70’s.
Factual memory is unanimously considered to
refer to memories whose natural expression involves a clause of the form
“remember that p” where p stands for a proposition which has the property of
being true or false[3].
But factual memory, according to the grammatical distinction, is not limited to
general knowledge, such as “I remember that Paris is the capital of France”. It
can also refer to personal events. When it refers to personal events, it would
be the equivalent of remembering not the past experience, but that something
personally experienced occurred or existed, as in the following example: “I
remember that I went with my family to Barcelona for the Easter holidays”.
Event memory, on the contrary, which has also been named “personal memory” or
“experiential memory”, is characterized by the gerundival construction in -ing: I remember verb + -ing[4],
as in “I remember reading your book” or “I remember the sun going down over the
Indian Ocean”. Other authors consider that nominalizations of verbs are also
possible objects of event memories, as in “I remember the hike through the
Berkshires”[5].
This is the most common way in which grammar has
been used as a criterion to distinguish two different kinds of memory.
Nonetheless, more recently, Bernecker[6]
broadened and reformulated this dual distinction in order to add new memory
categories. According to his grammatical taxonomy, propositional memory refers
to “any ‘substituent’ of the schema ‘S remembers that p’, irrespective of
whether ‘p’ refers to something one has personally experienced”[7]
or not. In addition, propositional memory can also take the form of “wh-clauses”, such as “I remember ‘who’, ‘whom’, ‘what’,
‘where’, ‘when’ or ‘why’”. These cases of extroversive propositional memories
contrast with cases of introversive propositional memories, which refer to
one’s own mental states, are necessarily in the first-person mode and contain a
second-order subordinated proposition with a verb in a past tense: “I remember
that I [remembered, thought, believed, inferred, etc.] that p”.
Non-propositional memory, on the other hand, not only refers to event-memory (I
remember verb + ing), but also to object-memory (I
remember + noun) and property-memory (I remember + noun + of noun). To
conclude, in Bernecker’s taxonomy the grammatical
criterion is completely detached and independent of the content one: it is the
sole syntactic form that determines the nature of a memory.
Nonetheless, the grammatical criterion does not
constitute a good criterion to distinguish memory kinds because it merely
provides just that: a grammatical distinction, while pretending to provide
something more than a grammatical distinction, such as a distinction of
psychological memory kinds. First, the grammatical criterion on one hand
divides memories that seem to have more properties in common and could be
explained by the same causal mechanism, such as “I remember that I read your
book” and “I remember reading your book”, both of which refer to a past event
experienced by the rememberer, and on the other hand
clusters together memories that do not share many properties, such as “I
remember that I read your book” (that + event experienced) and “I remember that
Paris is the capital of France” (that + fact or semantic knowledge). This
suggests that the grammatical criterion is exclusively based on syntax and does
not even consider semantics, which would be essential to the understanding of
memory, even from a grammatical point of view. Second, this syntactic
distinction is not even a good syntactic distinction: it is exclusively based
on the analysis of English and does not consider the fact that some other
languages, such as Japanese, do not have the same kind of gerundival constructions[8].
But the essential problem is that the grammatical criterion is intended to be
more than just a syntactic and semantic distinction; it presumes to assert a
difference at the psychological level, in the sense that it aims to explain
different ways in which we can remember and not different syntactic
constructions of the verb “to remember”. Maybe the grammatical criterion was
thought by some authors as a simple way of using different terms to express
different kinds of memories based on another criterion, such as the content
criterion. “Remembering that p” would express memories of facts or knowledge
whereas “remembering + ing”, memory of experiences.
If it is the case, the grammatical distinction should get stuck to its origins
and be simply used as a way of naming different memories in English, but it
should not be considered as an independent criterion to distinguish different
kinds of memories.
The vehicle criterion
By vehicle, I refer to the more known notion of
vehicle of representation or mean of representation in opposition to the notion
of representational content. The vehicle criterion is the oldest criterion to
distinguish between different memory kinds that can be found in the literature.
Although there are some references to this distinction in Plato’s writings, it
is Aristotle who first made it explicit in De Memoria et reminiscentia.
In this text, Aristotle distinguishes between memory (mneme) and
recollection (anamnesis). Aristotle considers that memory has a
particular affinity with imagination (phantasmata).
Memory is an affection (pathos), and imprinted trace, and thus refers to
the capacity of “having” images that correspond to previous sense perception,
knowledge or judgment. Recollection, on the other hand, is an organized search,
and thus, it presupposes an orderly method, usually called association of
ideas: the rememberer has to move through a series of
images that follow an order in a particular succession. It may be interpreted
that for Aristotle “memory is nonpropositional and is
similar to scenic memory”, whereas “recollection is propositional and is
similar to narrative memory”[9].
Although Aristotle’s distinction between two kinds of memory: mneme and anamnesis,
is not solely built on the mean of representation, but also in the underlying
processes: passive versus active, involuntary versus voluntary, it is generally
taken as an example of a distinction based on the vehicles of representation
especially because of its influence during the Antiquity and the Middle Ages.
Cicero, Plotinus, Augustine, Avicenna, Averroes, Albert the Great, Thomas
Aquinas, mostly all philosophers after Aristotle adopted the dichotomy between
a sensible memory that would be a kind of mental picture that reduplicates in
the mind past perceptual and sensory experiences, and an intellectual memory, a
recollection of ideas that is presented discursively as an argument or speech.
This
first distinction between visual memories and propositional memories does not
really parallel the intuitive distinction between memory of experiences and
memory of events. Nonetheless, someone could think that memories of experiences
are imagistic and memories of facts are propositional. Michaelian[10]
does not propose a specific account on the topic, but mentions in his book[11]
that semantic memories have propositional content, whereas episodic memories
consist primarily of imagistic content (although it may also include
propositional content). Whereas visual and spatial imagistic contents may be
considered to be essential to episodic memory—such as Rubin[12]
has defended—, spatial and visual imagery is certainly not sufficient to
distinguish episodic from semantic memory. Semantic knowledge or memories of impersonal
information of the world, such as “The Amazon is currently burning at a record
rate”, can also been retrieved through more than just propositional vehicles:
they can be constructed from different pieces of information that come in
different formats. For example, this memory can come with visual images of the
forest burning I saw on TV, of a map of Brazil or South America, etc. The
emotion system can also intervene in the retrieval of semantic memories: I can
remember the current Amazon rainforest fires with rage or sadness (although in
this case these emotions are not really part of the memory but show the way
these memories are appraised). On the other hand, although impairments of
visual imagery have been associated with amnesia[13],
visual and spatial imagery may not be really necessary to episodically
remember: blind people certainly remember, mainly through auditory imagery and
narrative forms[14].
In conclusion, episodic memories probably present a richer imagery than
semantic ones, but the presence of mental images, or of any other vehicle of
representation, is not in itself sufficient to make a distinction between
different kinds of long-term individual memory representations, and may not
even be necessary to characterize the particularity of one kind of memory
representation such as memory of events personally experienced.
In fact, many authors have proposed that there is a
single vehicle for all long-term individual memory representations, such as
Bartlett[15]
and Russell[16],
who considered that genuine and pure memories are pictures of past occurrences,
that is, memory-images. But more recently conceptualizations of episodic memory
have highlighted quite the opposite characteristic: episodic memory is
constructed from information carried by different vehicles. Probably the first
proponent of this conception was Paivio, who in the
1970’s reintroduced the notion of images as a particular kind of representation
different from the linguistic one. He postulated the dual code theory of
memory, which considers that visual and verbal information are encoded in two
systems that interact but are functionally independent[17].
Rubin[18]
also proposed a multimodal and multi-system model of episodic memory. According
to his model, different aspects of episodic memories are stored not in an
abstract format but in different systems, that is why episodic memories are
always formed by the mutual coordination of independent systems. These basic
systems are independent because each one uses different structures and
processes for fundamentally different kinds of information. Each one has also
its own functions, neural substrates and types of errors that affect memory.
These basic memory systems that provide different kinds of information are
generally coordinated by the explicit memory system and the search and
retrieval system, both of which are considered as the behavioral and neural
basis of episodic memory. So according to Rubin, the main system of episodic
memory does not encode specific information but coordinates information coming
from systems that are not memory-specific. Concerning these basic systems that
are not memory-specific, Rubin distinguishes different sensory systems, the
language and narrative systems and the emotion system. In relation to the
sensory systems, he considers that each one provides different kinds of
information to the episodic memory: “different senses process information about
different properties of the environment (e.g. electromagnetic radiation,
vibration, pressure, chemicals that contact the sensory surface), using
different transducer mechanisms and neural networks that have different short-
and long-term relevance to different aspects of the individual’s behavior”[19].
Visual, spatial, olfactory, auditory, gustatory and tactile forms of imagery
not only rely at the subpersonal level on different
systems but also present specific properties and characteristics at the
personal level that make them different from each other as well as from other
different kinds of representations, like linguistic ones. In fact, according to
Rubin empirical studies are consistent with the idea that visual and multimodal
objects and scenes are stored both in the original sensory systems and in the
language system, more specifically, in the language that has been used at the
time of the event[20].
Nonetheless,
the idea that memories are constructed from different bits of information
coming from different systems does not entail that the sources of these
different kinds of information are exclusively memory systems. As a matter of
fact, these different bits of information not only contribute to the
construction of memories but also to the construction of imaginative scenarios
and fictitious experiences, and the simulation of future episodes. Consider the
visual and spatial system. Research has shown that the generation, maintenance,
and visualization of complex spatial contexts, that is, the process of scene
construction, is a common process underlying episodic memory and imagination[21].
Patients with long-term visual memory loss cannot draw objects or describe
property objects from memory or have a visual image of objects upon
introspection[22].
But like blind people, they can certainly remember experiences from their past,
as well as other impersonal information about the world, in a linguistic or
other kind of sensory form. Therefore, these bits of visual and spatial
information that have been encoded in the visual and spatial system are not per
se different kinds of memories but just different kinds of information that
contribute to the construction of different kinds of mental states, such as
memories, fictional imaginings, future self-projection, navigation, etc. The
same could be said about the other systems, such as the emotion system, which
can also intervene in current perceptions, perceptions of fiction (watching a
movie, reading a novel), imaginings, future simulation, memories, just to give
some examples.
In conclusion, the vehicle criterion does not
constitute a good criterion neither to characterize memory nor to distinguish
different memory kinds. If the vehicle criterion was employed as a criterion to
distinguish different kinds of memories, it would be too broad and inclusive
and would not only include other mental states and experiences that are not
past-oriented and have a completely different nature (imaginings, simulations,
perceptions, etc.), but would not even do its job of distinguishing different
long-term individual memory representations. That is why it is better to
consider the vehicle criterion as a criterion to distinguish different kinds of
information that are used by different mental capacities, such as imagination,
future projection, navigation, etc., as well as in the construction of
different kinds of memories, such as memories of episodes personally
experienced in the past and memories of impersonal information about the world.
The causal criterion
The causal criterion to distinguish memories is
based on the previous condition that is necessary to remember a past event, and
also appears in relation to the philosophical distinction between memory of
facts and memory of events, as I show below.
Although the causal criterion may be at first
sight associated to the causal theory of memory, the major proponents of this
theory[23]
did not use it to establish a distinction between different kinds of long-term
individual memory representations. Bernecker adopted
a grammatical criterion to make distinctions and applied the causal theory to
all cases of propositional memory. Concerning Martin & Deutscher,
while at the beginning of their seminal paper, they suggest that there are
memories that require a previous experience of what is remembered and other
that do not, they finally analyzed both of them within the same causal
framework. In fact, a memory of some information gained through testimony does
not imply a direct experience of the event remembered but it does imply a
previous (and direct) experience of learning about that event through visual or
auditory perception. Because the causal condition comprehends memory of events,
general knowledge, and even procedural memories[24],
it is not useful per se to establish distinctions between different kinds of
memories.
This conclusion can be questioned if memories
are produced by causes that are different in nature. Different causes produce
in principle different memory phenomena. Some authors have proposed that memory
of events only requires as a previous condition that the remembered event is
based on a previous experience or perception of that event, whereas knowledge
or factual memory requires that the present knowledge or belief about a past
event is based on a past knowledge or belief. The introduction of the notions
of “knowledge” and “belief” refines the characterization of the kind of
“experience” that is necessary to remember events learnt through testimonial
sources (understood in a broad sense) in opposition to events that have been
directly experienced by the rememberer (realization,
participation or direct perception). Malcolm[25]
is an advocate of the strong sense of this criterion. He defines event memory
or personal memory in the following terms: “A person B personally remembers
something, x, if and only if B previously perceived or experienced x and B’s
memory of x is based wholly or partly on his previous perception or experience
of x”[26],
whereas factual memory is defined as “A person B remembers that p if and only
if B knows that p because he knew that p”[27].
Fernandez[28],
on the other hand, defends a weaker version: whereas factual memory of the
proposition p requires previous belief that p, event memory of p is possible
even if the rememberer had never believed that p
before. Despite this distinction, Malcolm[29]
for example has admitted the existence of some relation between these two kinds
of memories. According to him, personal memory entails a factual memory version
of itself: “A person who personally remembers the burning of the city hall
could be said, in this derivative sense, to personally remember that it burned”[30].
He named this memory personal factual memory in opposition to non-personal
factual memory. Here we have two possible interpretations: either Malcolm
considers that when we lose the capacity to bring back details of a past
experience the personal memory becomes a factual memory, or he believes that
while experiencing an event we always gain and form some knowledge, which is in
certain way encoded independently of the experience itself and can be later
recalled in the form of personal facts.
In any case, the causal criterion implies
epistemological assumptions about the way in which our memories are justified:
factual memories always require a prior belief, if not knowledge. This thesis
corresponds to the epistemic theory of memory. I will nonetheless avoid a deep
analysis of the epistemic theory of memory and just focus on the two
formulations mentioned earlier. For my purposes the only question that matters is if the criterion is useful to distinguish between
different kinds of long-term individual memory representations. It has been
criticized[31]
that factual memory is an umbrella category that includes two different kinds
of memories, i.e. memory of events personally experienced and memory of events
learned through testimony[32].
This criticism could be overcome if we take into consideration the distinction
made by Malcolm between personal and non-personal factual memories.
Nonetheless, this distinction would cluster together factual memory and
personal factual memories, which are memories that have little in common except
the fact of being caused by a past belief / knowledge, and would divide
personal memories in two different kinds for the only reason of being produced
or not by a past belief / knowledge. Furthermore, the most important criticism
points to the fact that a previous belief or knowledge is neither sufficient
nor necessary for having a factual memory. First, remembering—even factually
remembering—cannot be equated with knowing that or believing that. Cases of
relearning, in which a subject knows or believes that p at
time t1, then forgets p at time t2, and then relearns
p again at time t3, prove that someone can know or believe that p without for
that remembering that p[33].
This example shows that previous knowledge or belief is not sufficient for
factual memory and that factual memory cannot be equated to present belief or
knowledge. Secondly, previous knowledge and belief are not even necessary to
remember something. Cases of inattentive remembering, that is, when we are
inattentive and do not form a conscious belief while experiencing the event but
we form it later while remembering the event[34],
or cases when we do not consciously form a belief about p but then a defeater
of p is removed and thus we remember p[35],
or negative memories, like remembering not having done something, are common
counterexamples that have been used in the literature to argue that past
knowledge and beliefs need not be a necessary requirement to remember beliefs
or past experiences[36].
Therefore, previous knowledge or belief cannot be a criterion to identify
different kinds of memory: not all current beliefs and knowledge are memories,
and memories do not require previous knowledge and beliefs.
In consequence, it seems that the causal
criterion that distinguishes different kinds of memory in function of their
cause: previous perception/first-hand experience for event memory versus
previous knowledge[37]
or previous belief[38]
for factual memory, does not really constitute a good criterion to discriminate
what it pretends to discern: memory of personal experiences and memories of
information acquired through testimony.
However, there is another possibility to
establish different kinds of memories based on the cause. Instead of focusing
on the distal cause of memories, a distinction can be made in relation to their
proximate cause: the memory trace. Memories of events personally experienced
could be caused by different memory traces than factual memories. This is the
proposal done by Chen & al. who consider that the episodic memory trace is
one of the key elements that enables the construction of episodic memories.
Episodic memory traces refer to information stored in the brain about the gist
of one specific experienced event, and contrasts with traces that refer to
information of “multiple, personally experienced or not, episodes (semantic
information)”[39].
Nonetheless, Chen & al. admits that through repeated reactivation in the
construction of episodic memories, episodic traces can in certain way become
semantic[40],
and they even acknowledge that “there may be no clear line that divides
episodic memory traces from semantic information, since some cases cannot be
easily recognized as either one or the other, leaving the possibility that the
two are different by degree rather than by kind”[41].
Chen & al.’s model of episodic memory is more complex than that (it is not
only based on a memory trace), and a real criticism lies outside the scope of
this article. But it is worth it to mention that an episodic memory trace is
not sufficient to make a distinction between different kinds of memories. As I
will better develop on the section about the content criterion, first, it is
not clear how specific must be the event from which the episodic memory trace
derives. And second, in some cases semantic information can be tied to a memory
trace of a quite specific and single episode, such as when I remember an event
I have not first-hand experienced but that I have recently heard about on the
news. I can even forget the context in which I learnt that information, without
this implying that the memory retrieved is not caused by a single and
near-experience memory trace. Acquisition of semantic information may depend on
the acquisition of the episode of which they are a part[42],
and thus episodic memory traces may also be involved in the retrieval of
semantic information about the world.
Tulving’s serial-parallel-independent (SPI)
model of memory also considers that there are different memory traces that are
at the origin of different memory representations. According to the
serial-parallel-independent model, information about an event is processed and
encoded first in semantic memory and then in episodic memory, stored in
parallel in both systems and retrieved independently. Unlike Chen & al.,
the difference does not reside on the level of generality of the information
carried by the trace, but on the nature of the information encoded, which is
determined by the nature of the original information processed and the property
of the system in question[43].
So for Tulving there is no single trace or single
engram for an event (even for an experience-near event), but “different kinds
of information, representing the many different aspects of the event, are
stored at different independent storage sites”[44].
For example, the semantic memory system can process and encode the concepts in
relation with their meaning; its output reaches then the episodic memory
system, which “computes the temporal-spatial-contextual coordinates of the
incoming information in relation to already existing episodic information, or
to the self”[45].
Nonetheless, it is not really clear what are the particularities of each kind
of information, especially of the information carried by the semantic memory
trace. On the other hand, if the episodic memory trace refers to the
temporal-spatial-contextual information of an event, it faces significant
challenges, as it will be shown in the following section. What is more, the SPI
model has received criticism from some authors who questioned the idea that
information can only be processed and encoded in episodic memory via semantic
memory[46].
But independently of the more technical problems
related with the SPI model, and more broadly with the episodic memory trace, it
is possible to conceive that what causes different long-term individual memory
representations is the existence of different memory systems. Memory systems
are defined by Tulving as “organized structures of more elementary operating
components. An operating component of a system consists of a neural substrate
and its behavioral and cognitive correlates. Some components are shared by all
systems, others are shared only by some, and still others are unique to
individual systems”[47].
This means that episodic memory is an independent memory system that differs
from other memory systems such as the semantic memory system in the information
it processes, the laws and principles of its operations, its behavioral and
cognitive functions but also its neural substrates and mechanisms[48].
The notion of multiple memory systems is in general widely accepted[49].
Although it has been questioned by researchers who defend the idea that
episodic and semantic memory can be distinguished focusing on processes and not
on systems[50],
Tulving[51]
has denied the existence of an opposition between memory systems and memory
processes or mechanisms and has advocated for a complementarity of approaches
(specially through the development of his SPI model). The discussion of the
problems surrounding the notions of memory systems and memory processes are
beyond the scope of this article. But the point I would like to highlight is
that establishing different neural memory systems does not solve the problem of
distinguishing different kinds of long-term individual memory representations:
more needs to be said about the main operational and computational
characteristics that accompany different neural substrates, that is, about
those “components (…) that are unique to individual systems”.
In fact, Tulving was not foreign to this
concern: his own definition of system is not reduced to a neural system but
encompasses different operating components. Although in some writings he
suggested that there are a large number of characteristics that distinguish
episodic from semantic memory without specifying which ones are the main ones,
in other writings he suggests quite the opposite: the content of memories, and
in later reformulations their phenomenology, are presented as the defining
traits of each kind of long-term individual memory representation. In the next
two sections, I explore these two main characteristics that, as I show, are not
exempt of problems.
The content criterion
The content
criterion—a criterion based on the intentional object of the memory—is probably
the more intuitive criterion to establish different long-term individual memory
representations. This classic criterion has been widely used in philosophy:
many philosophers from the beginning of the 20th century have distinguished
between memory of events personally experienced (event-memory, personal memory
or experiential memory) and memory of facts (factual memory). As it has been
explained in a previous section, some philosophers founded this distinction on
the grammar, and thus any memory of the form ‘I remember that p’ was considered
as a factual memory. But for other philosophers, the difference is exclusively
based on the content, that is, on the intentional object of the memory: whereas
event-memory refers to memory of events personally experienced by the rememberer, factual memory refers to memory of abstract and
impersonal information about the world that has been acquired through
testimony.
Nevertheless, it was probably in the psychological
field that the distinction between these two kinds of memory was better
characterized. Although Tulving is not the author of the notions of “episodic
memory” and “semantic memory”, it was the first who presented them together as
different and opposite kinds of declarative memory.
Originally,
Tulving developed the difference between the content of these two memories by
specifying the kind of information encoded by these two different systems.
Whereas semantic memory retains information about the meaning of words and
concepts and their interrelations, episodic memory retains “information about
temporally dated episodes or events and temporal-spatial relation among these
events”[52].
So three kinds of information are retained in episodic
memory: the what, where and when of some particular event
that happened in the past. That is why this kind of memory has been called www
memory by Clayton & Dickinson[53].
Whereas the where and when would correspond to the setting of an
event, the what, that is, the salient happening within the setting would
correspond to the focal element of the event[54].
Because it is the contextual information of the event which is essential to
episodic memory (when semantic memory encoded interrelations between concepts,
it also encodes the “what” of an event) this first characterization of
episodic memory has been considered as a contextual description[55].
As McCormack explained “the general idea is that whereas factual memory simply
involves retrieval of the fact acquired during a given learning episode, the
corresponding episodic memory would involve remembering something about the
specific learning episode itself, namely the context in which the fact was
acquired”[56].
If this criterion seems, in principle, adequate to specify the characteristics
that distinguish memories of events personally experienced from memories of
facts about the world, and it also accommodates some empirical data, it does
not actually constitute a good criterion to distinguish different kinds of
long-term individual memory representations.
First,
Tulving suggests that contextual information is stored with event information
in memory but he neither explained how, nor in what consists this specific
relevant contextual information. In fact, this last assumption presents some
difficulties that have been expressed by McCormack[57].
First, we are often able to correctly recall perceptual properties of events
without accurately recalling the temporal and contextual information. Second,
we are able to recall temporal and contextual information of events not
personally experienced; and finally, we are able to make temporal and
contextual judgments based on ways different from the retrieval of stored
pieces of contextual information. All these criticisms point to the fact that
temporal and contextual information is neither necessary nor sufficient to
define the episodic character of a memory. As McCormack wrote, “if it is argued
that the temporal-contextual information is often not accessible for retrieval,
it is not clear what is useful about introducing this type of information to do
the job of ensuring the specificity of episodic memories”[58].
If we do not ensure the specificity of episodic memory, there is no possible
distinction between two kinds of memory.
Second,
Tulving establishes that episodic memories are memories of occurrences of an
episode, that is, “an event that is distinctive and separate although part of a
larger series”[59]
that happens in a particular place at a particular time. It has a beginning and
an end and the interval is filled with some action which involves the rememberer, either as an actor or as an observer[60].
This suggests that the notion of ‘event’ corresponds to a quite minimal and
short time slice of experience and, thus for example, the memory of some
holiday trip would be too broad to be considered as an episodic memory but will
not fit either as a case of semantic memory which is conceived as “culturally- shared
general knowledge (including facts and vocabulary)” and is characterized by
being “detached from its context of acquisition and devoid of any subjective
sense of mental time travel”[61].
This leaves the memories of personal general events in a sort of conceptual
limbo: they do not actually meet the criteria to be considered neither cases of
semantic memories nor episodic memories.
Finally, as I have anticipated in the section about
the causal criterion, there is the problem of the specificity of the event of
an episodic memory. The idea that the event remembered must be experience-near
to count as an episodic memory certainly dismisses general and very abstract
events as possible objects of episodic memories, but it does not clarify how
much “experience-near” the event should be in order to be an object of episodic
memory. Given that an event can always be divided into other more particular
events, the memory of meeting someone for the first time, for example, (which
can be decomposed into looking into each other’s eyes, saying “Hi”, shaking
hands, introducing themselves) could be considered as too general to count as a
real instance of episodic memory, but probably nobody would agree in denying
the episodic status to this kind of memory.
The concept of autobiographical memory appeared in
certain way to fill up this void by proposing that memories of our personal
past have different levels of abstraction and generality, are hierarchically
organized in a nested structure, and are not necessarily episodic in nature.
Because this concept proposes a richer characterization of memories of our
personal past than the sole notion of episodic memory, it presents a solution
to the problem of the conceptual limbo of memories of not experience-near
events which is characteristic of Tulving’s dichotomical
model, while avoiding the first problem: autobiographical memories are not
defined in function of their location in space and time but according to the
level of abstraction of the event remembered.
Different
computational and psychological models of autobiographical memories were
developed from the 80’s by Kolodner[62],
Linton[63],
and Barsalou[64],
among others. But probably Conway’s[65]
model is the most known because of its several reformulations. Although each
model presents its own characteristics and particularities (which I cannot
develop here), all of them recognize the existence of different levels of
abstraction of events that can go from those that are experience-near to those
that are highly abstract and involve a period of our lives. That is why when
remembering events of our personal past, the content of our memories can be a
simple event: opening the car to go to the airport; a complex event: driving to
go the airport; a general event, that includes repeated events: weekends at the
beach, and single general events: trip to Malaysia; an extendure:
living in the white house in Newtown; or an entire life-time period: our
childhood.
Furthermore, two important principles guide most of
the autobiographical memory conceptualizations. The first one states that
generic knowledge about events and life-time periods is involved in the
construction and representation of all memories, even of simple and
experience-near events. That is why these memories are never a literally a
record of experience but are always integrated with this generic and specific
knowledge. It is Conway[66]
who explained with more detail the memory of simple events. A simple episodic
memory (SEM) is composed by an episodic element (EE) and a conceptual
framework. The episodic element corresponds to a fragmentary and summary
representation of experience that is the result of a
sensory-perceptual-conceptual-affective processing of external stimuli. But the
episodic element is always embedded in a conceptual framework that
contextualizes it, interprets it and thus gives to it a personal meaning.
Experiences are in general complex and involve more than a simple episodic
memory, which is why in general a SEM relates to other SEM in a broader frame
giving rise to a complex episodic memory (CEM). Whereas a lot of episodic
memories that we formed during the day are if not lost at least inaccessible, a
small portion of them become integrated with autobiographical memory knowledge
structures—general events, lifetime periods, self-knowledge, etc.—and this
integration enables them to remain accessible in long term memory.
The second principle, mainly highlighted by Neisser[67],
Barsalou[68]
and Conway & Pleydell-Pearce[69],
considers that in everyday memories rememberers move
through this nested structure, and so the contents of our personal memories are
not stable and punctual, but fluctuate from simple events to more complex and
abstract events and vice versa. According to Neisser, “recalling an experienced
event is a matter not of reviving a single record but of moving appropriately
among nested levels of structure”[70].
And in fact, empirical research done by Barsalou
confirms that in free recall protocols, the memories of simple events are
invoked together with other kinds of memories of events, like complex and
general events (“I went on a diet”) and alternative events (“I had not taken a
shower”), and even evaluative comments about aspects of the events (“We had a
lovely apartment”).
Whereas
this second version of the content criterion avoids the problems of the first
version, it presents some weakness as a criterion for distinguishing different
kinds of memories. On the one hand, the distinction between different memories
of personally experienced events according to the level of abstraction of
events (from experience-near to highly abstract events) is not really conceived
as a distinction per se but parallels the episodic/semantic distinction.
Shortly after its reintroduction in the field, autobiographical memory was
considered as a synonym of episodic memory[71].
But very soon it was distinguished from episodic memory, without for that being
defined and analyzed independently from episodic memory. Larsen[72]
conceived autobiographical memory as a subset of the episodic memory system,
Rubin[73]
and Baddeley[74]
as a subsystem at the intersection of episodic memories and semantic memories,
and Brewer[75]
and Conway & Pleydell-Pearce[76]
as a superordinate system that includes both episodic memory and semantic facts
about the self. However, this superposition or partial overlap between
autobiographical memories, episodic memories and semantic memories is not
really clarifying. For example, the conception that not all episodic memories
are autobiographical memories leads us to the following paradox: episodic autobiographical
memories would share a large set of properties with episodic
non-autobiographical memories in order for both of them to be considered as
part of the same memory kind, but at the same time, they would present a large
set of different properties from those belonging to episodic
non-autobiographical memories in order for them to be gathered together with
semantic autobiographical memories in another memory kind (i.e.,
autobiographical memories). The same could be said for semantic autobiographical
and non-autobiographical memories. What is more, it seems that what reunites
episodic and semantic autobiographical memories and distinguishes from other
episodic and semantic memories that are not autobiographical is not the nature
of the event, but some implicit criterion related to the reference to the self
(I will come back to this idea in the last section).
On
the other hand, considering that autobiographical memories of experience-near
events equate to episodic memories whereas the rest of autobiographical
memories constitute a subset of semantic memory[77]
does not provide a better solution with regard to the empirical data that shows
the heterogeneous nature of autobiographical memories that are not
near-experience. Even if it is granted that experience-near autobiographical
memories have many properties in common (similar level of abstraction and
generality, similar neural substrate, a high degree of sensory imagery,
specific phenomenal feelings, etc[78])
and thus it makes sense to cluster them into the category of episodic memory,
the same cannot be said of semantic autobiographical memories.
First,
memories of repeated events include spatial and temporal information, that is,
contextual information, sensory imagery and emotion[79],
similarly to episodic memory. Also, the brain areas activated during retrieval
are more similar—even if they do not overlap—to episodic memory than to
semantic memory: they include the hippocampus and the medial temporal lobe. The
phenomenal feeling of re-experiencing is also generally present in memory of
repeated events.
Secondly,
the nature of autobiographical facts, like “My son was born the 15th June 1999”
or “I used to live in Brown Street” is quite complex. Although some literature provide evidence of the similarity between autobiographical
facts and semantic memory[80],
a recent review article about amnesic patients with medial temporal lobe lesion
made by Grilli & Verfaellie[81]
comes to a different conclusion. According to this review, memory of
autobiographical facts is impaired in amnesic patients, and this shows that
memory of autobiographical facts partially depends on the medial temporal lobe
and thus, that they share mechanisms with episodic memory. Also, Martinelli
& al.[82]
state that semantic autobiographical memories seem to recruit basically the
same brain structures as does episodic memory, but to a lesser extent and
excluding the hippocampus. The problem is that notions like autobiographical
facts and semantic autobiographical memories seem to refer to a variety of memories
that includes friend’s names and personal addresses, memories that are closely
associated with a unique event, such as “My son was born the 15th June 1999”,
and memories that refer to lifetime periods, such as “I used to live in Brown
Street”. It is probable that memories of friend’s names and addresses are more
semantic—because of their frequency—than other kinds of memories associated
with events. It is also probably that memories associated with specific events
would include more spatio-temporal information,
sensory imagery and emotional content than memories associated with general
events or lifetime periods. However, in real life memories the content is not
always the same but fluctuates from the general to the specific and vice versa,
so it is also probable that even memories of general events and lifetime
periods can also be experienced with sensory imagery and some emotional
content, with a consequent feeling of remembering. To someone’s question as to
if I had always lived in the same house or not, I could simply reply “No, I
used to live in Brown Street” and simultaneously a mental image of my house and
the street could come to my mind as well as some feeling of nostalgia.
There
is also a third kind of autobiographical memories that are not experience-near
but present some similarities with episodic memories: autobiographically
significant concepts[83],
which refers to semantic concepts related to events (such as famous faces and
names). This kind of memories seem to involve the medial temporal lobe and so
engage both the episodic memory system and the semantic one. They are in
general characterized by a high degree of sensory imagery and emotional content
and include spatial and temporal information. Larsen[84]
also mentions a similar case to autobiographically significant concepts:
narrative memories. Narrative memories are memories about events not directly
experienced by the rememberer but learnt indirectly
by testimony, such as events learnt by the news or read in a book or told by
someone, whose content is thus not personal but, like autobiographically
significant concepts, are nonetheless autobiographical because they have the
property of being tied to a personal event. The personal context corresponds to
the moment in which the rememberer received the
message, news or information[85],
and this moment refers not only to the time and location but also can include
personal meaning and have an emotional tone. It would be expected that
narrative memories could be explained in a similar way to autobiographically
significant concepts, astride the episodic memory and the semantic one.
All
these cases point to the fact that autobiographical memories that are not
experience-near are very heterogeneous and cannot be clustered into the
category of semantic autobiographical memories, in opposition to episodic
autobiographical memories. Specific events, general events, repeated events,
lifetime periods and other kinds of personal memories can be understood neither
as similar to episodic memory nor as a sub-domain of semantic memory; nor even
as a continuum that goes from the most specific and episodic elements to the
most abstract and semantic ones, like most of the nested models of
autobiographical memories propose. Furthermore, some evidence indicates that
memories of experience-near events do not really belong to a homogenous
category, and thus cannot always considered to be episodic. In fact, there are
some memories of experience-near events that are poor in imagery and emotional
tone, do not depend anymore upon the MDL (and thus upon the hippocampus), and
present a large set of phenomenal properties that are different from those
generally attributed to episodic memory. The reason is the process of semantization that affects some episodic memories due to
remoteness and aging[86].
To
conclude, this second version of the content criterion does not offer a good
criterion to distinguish different kinds of memories neither. Similarities and
differences between autobiographical memories of events cannot be explained
according to the level of abstraction and generality of the event remembered.
Although the level of abstraction of the content remembered may play some role
at the moment of distinguishing different kinds of long-term individual memory
representations, it does not provide by itself neither a sufficient criterion
nor a significant one. It seems thus that the nature of memories are quite independent of the nature of the content that
constitutes their intentional object, either when the content is defined in
function of the absence or presence of temporal and contextual information, or
in function of its level of abstraction and generalization.
The phenomenal criterion
Unlike the content criterion, which is focused on the
properties of the intentional object of memories in order to distinguish
different memory kinds, the phenomenal criterion establishes this distinction
according to the particular phenomenology of memories. With time, the
phenomenology of memory acquired more importance in Tulving’s memory model and
thus memory kinds were redescribed in phenomenal
terms: instead of being mainly defined in terms of the kind of information that
is encoded, stored and retrieved, they were redefined in terms of the kind of
subjective experience that accompanies the operations of the system at
retrieval. McCormack[87]
called this new way of defining episodic memory the experiential description.
This criterion proposes that the essential characteristic for identifying
memory kinds is the type of conscious awareness that accompanies a memory
experience. According to Tulving[88],
whereas the notion of autonoetic awareness characterizes episodic memory,
semantic memory is characterized by the notion of noetic awareness. The
concepts of noetic and autonoetic awareness, however, are not well defined. For
example, Tulving describes autonoesis as referring
“to the kind of consciousness awareness that characterizes conscious
recollection of personal happenings”[89].
This description is a tautological definition; it does not say what
characterizes the awareness of recollecting personally experienced events. In
fact, Tulving added to autonoesis another three
essential features of episodic memory which can give a clue about the meaning
of this term: mental time travel, self and chronesthesia,
all of which are highly related. Mental time travel refers to the possibility
for the rememberer “to travel back in his or her mind
to an earlier occasion or situation in the rememberer’s
life, and to mentally relive the experienced and thought-about happenings”[90].
The self refers to the agent who does the travelling, and chronesthesia
to the experience of subjective time that recollection allows. These
characteristics summarize the meaning of the particular consciousness that
accompanies recollection, and so autonoesis could be
defined as the awareness that arises from the ability to subjectively and
mentally time travel to the past, which allows the self to represent and
re-experience a past event. Therefore, essential to the notion of autonoesis is the idea of re-experiencing past experiences[91].
Because of these characteristics, it “confers the special phenomenal flavor to
the remembering of past events, the flavor that distinguishes remembering from
other kinds of awareness, such as those characterizing perceiving, thinking,
imagining, or dreaming”[92].
Noesis, on the other side, only allows the rememberer
to be aware of the content of the memory. When information about the personal
past is remembered without any recollective experience, that is, with noetic
awareness, it is strictly only known and corresponds to the semantic memory
system. That is why this distinction between remembering a past experience in
an autonoetic or a noetic manner corresponds to other famous distinction
introduced by Tulving in the psychological field between remembering some event
from the personal past and knowing some event from the (personal) past.
Some amnesic patients would constitute empirical
evidence for this dissociative hypothesis: those who have source amnesia and
those with medial temporal lobe damage which principally affects the
hippocampus. Source amnesia is described as the retention and retrieval of
factual content of an episode without any recollection of the episode itself in
which that factual content was acquired[93].
Patients with frontal lobe damage, elderly individuals and patients in early
stages of Alzheimer’s disease are among the groups who have this sort of memory
disorder, which was already remarked by Claparède[94]
in the study of his Korsakoff's patient. The same dissociation is presented in
amnesic patients with damage to the hippocampus. It is of interest to consider
patients whose hippocampus has been damaged early in life[95],
because it shows that they can attain average levels of speech, language
competence, literacy and factual knowledge without being able to remember
events in daily life and spatial and temporal locations. But probably the most
important and paradigmatic case of loss of autonoetic consciousness without
loss of noetic consciousness is the amnesic patient K.C.[96]
who had global anterograde amnesia and retrograde episodic amnesia. K.C.’s
implicit and procedural memories were preserved. He still possessed stored
semantic facts about the world and also about his life, and could still learn
under special conditions some new facts. However, he was unable to recollect
specific events in which he himself participated or that he witnessed, whether
experienced long ago or in more recent times. In fact, the autobiographical
knowledge about his past relied completely on his semantic memory: he showed no
feeling of re-experiencing and his recollections completely lacked the
subjective re-evoking of the emotional and contextual details that defined
episodic memory[97].
Even episodes in K.C.'s life that are supposed to be personal tragedies are
remembered as simple decontextualized facts, without any emotional tone,
vividness or intensity. K.C. is not the only example and other similar cases
have been reported in the literature[98].
More
recently, the psychologist Stanley Klein[99]
has defended a similar position to Tulving’s experiential definition episodic
memory. According to Klein, “a memory experience is not the nature of the
content presented to awareness, but the manner in which awareness becomes
associated with the content during the act of retrieval”[100].
Information about an event’s what, where and when, that is, about the temporal
and spatial context of some happening, even if it is self-referential, can also
be displayed in reports of semantic experience. Unlike Tulving’s SPI model,
Klein considers that this information is unspecified prior to its demarcation
as semantic or episodic during the act of retrieval. Two case studies are in
favor of his account. The first one refers to amnesic patients similar to K.C.,
as well as developmental amnesic children, who are unable to re-live their
personal past, and thus remembering it in a proper way, but are nonetheless
able to re-learn the temporal, spatial and self-referential information about
their own past even if less detailed than non-amnesic people. The second case
is a little more controversial, because it refers to a patient who had a
temporary memory disorder due to an accident and was exclusively interviewed at
that time by Klein. According to Klein & Nichols[101],
during some time after the accident this patient, who is known as R.B., was
able to describe events of his life with contextually rich details but without
the feeling of mine-ness that would be proper to autonoetic awareness: “when I
recall memories of my past I intellectually know they are about me. It just
does not feel like it ... when I remember scenes from before the injury they do
not feel as if they happened to me –though intellectually I know they did”[102].
In
conclusion, the criterion based on the phenomenology of the experience breaks
the equation between episodic memory and memory of events. The retrieval of
information about some experience-near past event does not need to be
necessarily accompanied by autonoetic awareness; it can also come with noetic
awareness. What is more, the memory distinction based on the phenomenology can
account without problem what the second version of the content criterion could not:
the process of semantization of memories of
experience-near events due to remoteness and aging[103].
The
phenomenal criterion presents other advantages in relation to the second
version of the content criterion. In general terms, it avoids multiplying different
memory systems and kinds and thus analyzing a single memory occurrence as
belonging to or sharing properties with different memory systems (such as
semantic autobiographical memories). It also avoids clustering under the same
category: semantic autobiographical memory, all the memory of events that are
not experience-near, which do not really belong to a homogenous class (as the
empirical data concerning repeated events, autobiographical facts and
autobiographical significant concepts have shown). Memories of events can be
remembered autonoetically or noetically,
and this is not completely determined by the level of abstraction and
generality of the event remembered. That is why a memory of a repeated or
general event can be remembered with autonoesis, and
thus present more similarities with some memories of experience-near events
than with other memories of events of a high level of abstraction. It is in
fact the mode of retrieval which determines the kind of awareness that
accompanies the memory retrieval. Finally, this criterion does not distinguish
different kinds of memories of events, but different kinds of memories tout
court. So it also accounts for memories of
individuals, that is, of objects, people and places, which are not actually
reducible to events and were in general excluded from most models of
autobiographical memory. It explains thus why the memory of my partner who is
currently absent from home, or of my dead father, can come with a subjective
feeling of travelling back to the past.
Nonetheless,
the phenomenal criterion is not exempt of criticism. First, as I have already
highlighted, the notions of autonoetic and noetic awareness are quite obscure.
Since its introduction by Tulving, autonoesis has
been considered by many researchers as synonym of reexperiencing and
consciously reliving the past. Authors insist that in episodic memory the rememberer “view[s] these events as they would originally
have been seen through her own eyes”[104],
and it “is for her just as if the scenario were present”[105].
Nonetheless, these characterizations are rather vague and imprecise to
distinguish memory kinds. What is more, autonoesis,
or the feeling of reliving and reexperiencing the past does not appear as an
all-or-none phenomenon. Only traumatic memories related to PSTD are cases that
truly deserved the qualifying “reexperience” and “reliving”: the intrusive
visual memories that characterize PTSD convey a sense of immediate perceptual
experience because they are processed as real and as a real threat[106].
Other cases of episodic memories do not imply a real feeling of reexpericing or reliving the past event. In fact, this
variability of the sense of reliving a past experience is taken into
consideration in different versions of the Autobiographical Memory Characteristics
Questionnaire that demand the rememberers to address
the vividness and the level of recollective experience of their memories using
a seven-point Likert-type scale[107].
Therefore, autonoesis seems to be better conceived as
a gradual phenomenal property of memories that can come in different degrees,
and not as a particular mode of awareness among two possible and opposite
modes.
Secondly,
although it is possible to conceive the phenomenal criterion as the key
criterion to establish distinctions between memory kinds, it is not clear
neither that Tulving and subsequent researchers conceived it independently of
the content remembered (except probably only for Klein), nor that it is
possible—and plausible—to conceive it that way. In one of Tulving’s writings,
episodic memory “is defined in terms of criteria such as the system’s
function—what the system does, how it works, the kind of ‘information’ that it
deals with, its relations to other systems, and its neural substrates”[108],
suggesting that the content of a memory plays some role in order to determine
the memory kind to which this memory belongs. An episodic memory, that is, an
experience remembered with autonoetic awareness, does not seem to be a memory
of any kind of information.
Rubin[109],
on the other hand, is less equivocal than Tulving: for him, memories that are
highly correlated with a sense of reliving and other measures of reliving (such
as mental time travel) are memories presented through a particular kind of
vehicle: visual and spatial imagery. Whereas he does not explicitly correlate
phenomenology and content but phenomenology and vehicles of representation,
this explicit correlation carries in certain way an implicit correlation
between content and phenomenology. The nature of visual and spatial imagery or
“scenes”, as Rubin calls it, puts certain constraints on the kinds of
intentional object that our memories may have. Because the construction of a
unique scene resultant from a sort of superposition form the exposure to
multiple perspectives of the same scene does not mainly differ from the
superposition of repeated scenes implied in the construction of scenes of
repeated events, Rubin considers that repeated events are similar to unique and
simple events[110].
But the relation between scenes and more abstract and generic events is not so
simple. In some cases it is possible to construct a
sort of generic image that does not correspond to any single episodic moment,
whereas in other cases a similar generic image is not possible:
Generic personal
memories are constrained by the abstracting properties of the relevant
perceptual systems. Thus, I can have a generic personal memory of ‘going out on
the beach’ during some vacation but not of ‘going on vacation’ where that includes
going hiking in the mountains, going swimming at a beach, and visiting a major
city[111].
So
it seems that when our memories are directed to some general events, extendures and lifetime periods, the only imagery that can
accompany these retrievals to give to them a sort of derivative recollective
phenomenology are scenes of less general events that are subsumed under them or
of things or people involved, that is, images of what can be representable
through the corresponding perceptual systems.
This
correlation between phenomenology, content and vehicles of representation
depends on the assumption that visual and spatial imagery is what gives a
special kind of phenomenology to memories. Nonetheless, this assumption is not
exempt of problems: scenes alone cannot account for the special phenomenology
that is characteristic of some memories because, as it was explained as part of
the criticism to the vehicle criterion, memories of specific events can be
devoid of visual and spatial imagery; scenes can also be part of memories with
noetic awareness, and visual and spatial imagery are also part of the
construction of different kinds of non-mnemonic mental states.
On
the other hand, there is certainly some correlation between the phenomenology,
content and means of representation of memories. Nonetheless, these possible
correlations have been for now more sketched that studied in deep, and in fact
seem difficult to define. Rubin’s proposal of a relation between vehicles of
representation and memory contents conceived according to its level of
abstraction and generalization present some limitations. Although Tulving[112]
himself implied that there were multiple properties that establish a difference
between episodic and semantic memory, he did not explicitly explain how these
properties interrelate with each other in order to determine the ones that are
essential and the ones that are secondary. His later emphasis on the major
importance of the phenomenal property to define memory kinds (autonoesis versus noesis), does not seem to be completely
convincing.
A promising line for future
research: the affective criterion
This analysis shows that for now, none of the
criteria proposed in the literature to distinguish memory kinds is completely
satisfactory. Some criteria should be directly discarded, such as the
grammatical and causal criteria, because they do not provide any credible hint
of the memory properties that may be the cause of different natural kinds. Some
other criteria seem more promising, but neither as they were formulated nor
isolated from one another. The last remarks done in the former section suggest
that a kind of correlation between the vehicle of representation, the content
and the phenomenology may be better suited to account for the varied nature of
our memories. This interrelation, which is probably not easy to characterize,
needs certainly to be object of a detailed and deep research. It is not the aim
of this article to provide such an explanation: its main purpose is to expound
the different criteria proposed to distinguished memory kinds and highlight
their problems and challenges. Nonetheless, the rest of this section mentions
briefly a hypothesis that may be worth exploring.
Many authors, especially coming from philosophy,
have stated that what really characterizes episodic memories and distinguishes
them from semantic ones is the presence of the past self in the representation
of the experience remembered. Whereas in episodic memory, the events
represented are presented as having happened in the past to me, in semantic
memory, on the contrary, those events do not present any reference to the self.
This reference to the self has been in some cases tied to the notion of autonoesis: autonoetic awareness is not only awareness
directed to something past, but also directed to onself,
more specifically, to the past self who was the subject of the past experience.
For example, Chen & al. have explicitly defined autonoesis
as “the awareness that oneself has been part of the recalled event”[113];
also Bermudez[114]
has stated that episodic memory involves an awareness that the event happened
to me, to give some examples. This self-reference can be accounted by two
means. Either the self is directly present on the memory content (for example,
through the integration of a memory or conceptual knowledge of my past self
into the construction of the memory of a past experience), or is part of the
process of retrieval (option that would be more compatible with memory
processes than with memory systems). Rowlands[115]
has recently proposed this last conception: the act of remembering—which can be
understood as the process of remembering—subsumes the event remembered under a
specific mode of presentation that, as a result, entails a change in the way
the event is remembered. Because of this mode, the past event is presented “as
one that has been formerly witnessed, orchestrated or otherwise encountered by
the rememberer”[116],
that is, in intrinsic connection to the past self.
Whereas the line of research that proposes the presence
or the absence of the self as a criterion to distinguish different memory kinds
is quite promising, it needs further development in order to avoid a similar
criticism to the previously mentioned, such as the semantization
process suffered by some episodic memories due to remoteness and aging.
One possible way of refining this idea and avoid
this possible criticism, consists in exploring a more elaborated and precise
way in which the self can be present or not in our memories: through affect.
Affect is essentially relational, that is, is the product of the relation
between the individual and her environment, and thus represents how an
individual is moved and influenced in terms of harms, benefits, morality or
self-image by the dynamic and continuous environmental changes. While some of
our memories seem to be directed towards objective and impersonal properties of
past events and experiences, others represent not the past experience or event
in itself, independently of the rememberer, but the
way in which the rememberer was affected by that past
experience or event. The focus of this last kind of memories is the affective
significance of a past experience, and not its descriptive impersonal
properties. This affection can be explicitly attended to as the intentional
object of the memory, or can appear in a pre-attentive or pre-reflective way;
it can take the form of interoceptive bodily sensations, action tendencies, and
even motor behaviors. The interplay between a special type of memory content:
affect, carried by specific vehicles of representation: bodily sensations,
action tendencies, etc., and being susceptible of different levels of
consciousness (reflectivity, pre-reflectivity, etc.), would determine the
particular phenomenal properties of the memory. According to this proposal, our
everyday memories are situated somewhere between two extremes: either we
remember events and experiences objectively, completely detached from the rememberer, or we remember them through their past
affection, and it is through this affection that the rememberer
is present in the content of memory. Close to the first extreme, there are
memories of past events that contain a minimal reference to the self: events
are presented as formerly witnessed, orchestrated or encountered by the rememberer, so the presence of the self in the memory
content is reduced to a simple past spectator or orchestrator. At the end of
the opposite extreme, there are those “affective” memories that imply a real
emotional reexperience of the past event, such as in PTSD. Because of the
varied way in which affection can be represented and consciously attended to,
our affective memories would occupy different points in this continuum, and
entail different phenomenal properties, or similar phenomenal properties but in
different degrees. In fact, this conceptual outline could be matched with
relatively recent attempts to integrate the amygdala to the standard model of
episodic memory in order to account for the emotional properties and
significance of the events remembered through the episodic memory system[117].
More research should be pursued in this line,
because the presence of the self through affection appears at first sight as an
excellent starting-point in order to identify the possible interrelations between
key components that account for the existence of different kinds of long-term
individual memory representations.
Acknowledgements
This research was mainly supported by a grant
funded by the Fondation Maison des sciences
de l’homme and the Centre Universitaire
de Norvège à Paris (France). It was presented at
the Centre for the Study of Mind in Nature of the University of Oslo (Norway)
in September 2016. I would like to thank to the audience, especially to Olav Gjelsvik, Sebastian Watzl and Frøydis Gammelsæter for their
feedback.
References
Allen,
P. A., Kaut, K. P., & Lord, R. R. Emotion and episodic memory. In
DERE, E., EASTON, A., NADEL, L., & HUSTON, J. P. (Eds.), Handbook of
episodic memory. Handbook of behavioral neuroscience. Elservier,
vol. 18, 2008, p. 115-132.
AMBERBER,
M. (Ed.), The Language of Memory in a Crosslinguistic Perspective.
Amsterdam: John Benjamins Publishing Company, 2007.
ARISTOTLE
(350 BC) On Memory and Reminiscence.
Baddeley,
A. Reflections on autobiographical memory. In: BERNTSEN, D. & RUBIN, D.
(Eds.). Understanding
Autobiographical Memory. Cambridge: Cambridge University
Press, 2012.
Barsalou, L. The content and organization of autobiographical
memories. In: NEISSER, U. Neisser & E. Winograd (Eds.), Remembering
reconsidered. Cambridge: Cambridge University Press, 1988.
BARTLETT,
F.C. Remembering. Cambridge: Cambridge University Press, 1932.
Bermudez,
J. L. Memory and Self-consciousness. In: BERNECKER, S. & MICHAELIAN, K. (Eds.).
The Routledge handbook of philosophy of memory. New York: Routledge, p.
180-191, 2017.
BERNECKER,
S. The Metaphysics of Memory. Dordrecht: Springer, 2008.
BERNECKER,
S. Memory: A philosophical Study. Oxford, Oxford University Press, 2010.
BLAXTON,
T. Investigating dissociations among memory measures: support for a transfer appropriate processing framework. Journal of Experimental
Psychology: Learning, Memory and Cognition, 15, 657-658, 1988.
BOYACIOGLU,
I., & AKFIRAT, S. Development and psychometric properties of a new measure
for memory phenomenology: The autobiographical memory characteristics
questionnaire. Memory, 23(7), 1070-1092, 2015.
Boyd,
R. Homeostasis, species and higher taxa. In: WILSON, R. (Ed.), Species:
New Interdisciplinary Essays. Cambridge, MA: MIT Press, 141-185, 1999.
Brewer,
W. What is autobiographical memory? In: RUBIN, R. (Ed.) Autobiographical
Memory. Cambridge: Cambridge University Press, 25-49, 1986.
CABEZA,
R., & Moscovitch, M. Memory systems, processing
modes, and components: functional neuroimaging evidence. Perspectives on
Psychological Science, 8(1), 49-55, 2013.
CHENG,
S., WERNING, M., & SUDDENDORF, T. Dissociating memory traces and scenario construction
in mental time travel. Neuroscience
& Biobehavioral Reviews, 60, 82-89, 2016.
CLAPARÈDE,
E. Recognition et moitié. Archives de psychologie, 11, p. 79-90, 1911.
CLAYTON,
N.S. & DICKINSON, A. Episodic-like
memory during cache recovery by scrub jays. Nature, 395, 272–274, 1998.
Conway,
M. A structural model of autobiographical memory. In: CONWAY, M. RUBIN, D.,
SPINNLER, H., & WAGENAAR, W. (Eds.). Theoretical perspectives on
autobiographical memory. Dordrecht, Kluwer Academic Publisher, 167-194,
1992.
Conway,
M. A. Autobiographical knowledge and autobiographical memory. In: RUBIN, D.
(Ed.). Remembering our past: Studies in autobiographical memory.
Cambridge: Cambridge University Press, 67-93, 1995.
CONWAY,
M. Memory and the self. Journal of Memory and Language, 53, 594-628, 2005.
CONWAY,
M. Episodic memories. Neuropsychologia, 47,
2305-2313, 2009.
CONWAY,
M., & PLEYDELL-PEARCE, C. The construction of autobiographical memories in
the self-memory system. Psychological review, 107, 261-288, 2000.
DOLAN,
R. J., LANE, R., CHUA, P., & FLETCHER, P. Dissociable temporal lobe
activations during emotional episodic memory retrieval. Neuroimage,
11(3), 203-209, 2000.
EVANS,
F. & THORN, W. Two types of posthypnotic amnesia: Recall amnesia and source
amnesia. The International Journal of Clinical and Experimental Hypnosis,
14(2), 162-179, 1966.
FARAH,
M. J. The neurological basis of mental imagery: A componential analysis. Cognition,
18, 245–272, 1984.
FERNANDEZ,
J. The intentionality of memory. Australasian Journal of Philosophy, 84(1),
39-57, 2006.
GHAHAM,
K.S., SIMONS, J., PRATT, K.H., PATTERSON, K., & HODGES, J.R. Insights from
semantic dementia on the relationship between episodic and semantic memory. Neuropsychologia, 38, 313-324, 2000.
GREENBERG,
D. L., & RUBIN, D. C. The
neuropsychology of autobiographical memory. Cortex, 39(4-5), 687-728,
2003.
GRILLIA,
M. & VERFAELLIE, M. Personal semantic memory: Insights from
neuropsychological research on amnesia. Neuropsychologia,
61, 56-64, 2014.
HASSABIS,
D., KUMARAN, D., & MAGUIRE, E. Using imagination to understand the neural basis of
episodic memory. The Journal of Neuroscience, 27, 14365-14374, 2007.
HOLMES,
E. A., & MATHEWS, A. Mental imagery in emotion and emotional disorders. Clinical
psychology review, 30(3), 349-362, 2010.
JOHNSON,
M. K., FOLEY, M. A., SUENGAS, A. G., & RAYE, C. L. Phenomenal
characteristics of memories for perceived and imagined autobiographical events.
Journal of Experimental Psychology: General, 117(4), 371-376, 1988.
KLEIN,
S. The two selves: Their metaphysical commitments and functional
independence. New York, Oxford University Press, 2014.
KLEIN,
S. What memory is? WIREs Cognitive Science, 5(2), 1-38, 2014b.
KLEIN,
S. & NICHOLS, S. Memory and the sense of personal identity. Mind,
121, 677-702, 2012.
KOLODNER,
J. L. Maintaining organization in a dynamic long-term memory. Cognitive
Science, 7(4), 243-280, 1983.
KOLODNER,
J. L. Retrieval and Organizational Strategies in Conceptual Memory: A
Computer Model. Hillsdale: Lawrence Erlbaum Associates, 1984.
LACKEY,
J. Memory as a Generative Epistemic Source. Philosophy and Phenomenological
Research, 70 (3), 636-658, 2005.
LARSEN,
S. Personal context in autobiographical and narrative memories. In: CONWAY, M.,
RUBIN, D.,
SPINNLER,
H. & WAGENAAR, W. (Eds). Theoretical Perspectives on Autobiographical
Memory. Dordrecht: Kluwer Academic Publisher, 53-74, 1992.
Linton,
M. Ways of searching and the contents of memory. In: RUBIN, R. (Ed.), Autobiographical
Memory. Cambridge: Cambridge University Press, 1988.
MACHERY,
E. Doing Without Concepts. Oxford: Oxford University Press, 2009.
MALCOLM,
N. Knowledge and Certainty. London: Cornell University Press, 1963.
MARGOLIS,
J. Remembering. Mind, 86, 186-205, 1977.
MARIAN,
V., & NEISSER, U. Language-dependent recall of autobiographical memories. Journal
of Experimental Psychology: General, 129, 361–368, 2000.
MARTIN,
C. B. & DEUTSCHER, M. Remembering.
The Philosophical Review, 75(2), 161-196, 1966.
MARTINELLI,
P., SPERDUTI, M. & PIOLINO, P. Neural substrates of the self-memory system. Human
Brain Mapping, 34, 1515-1529, 2013.
McCormack,
T. Attributing episodic memory to animals and children. In: McCORMACK,
T. & HOERL, C. (Ed.) Time and Memory. New York: Oxford University
Press, 285-313, 2001.
MICHAELIAN,
K. Opening the doors of memory: is declarative memory a natural kind? Wiley
Interdisciplinary Reviews: Cognitive Science, 6(6), 475-482, 2015.
MICHAELIAN,
K. Mental Time Travel: Episodic Memory and Our Knowledge of the Personal
Past. Cambridge, MA: The MIT Press, 2016.
MOSCOVITCH,
M. Memory and working with memory: A component process model based on modules
and central systems. Journal of Cognitive Neuroscience, 4, 257-267,
1992.
MUNSAT,
S. The Concept of Memory. New York: Random House, 1967.
NADEL,
L. & MOSCOVITCH, M. Memory Consolidation, Retrograde Amnesia and the
Hippocampal Complex. Current Opinion in Neurobiology, 7, 217-227, 1997.
NAYLOR,
A. Remembering-that: episodic vs. semantic. Philosophical Psychology,
24(3), 317-322, 2011.
NEISSER,
U. Nested structure in autobiographical memory. In: RUBIN, R. (Ed.), Autobiographical
Memory. Cambridge, Cambridge University Press, 71-81, 1986.
Nikulin, D. Memory in Ancient Philosophy. In: NIKULIN, D.
(Ed.), Memory: A history. Oxford University Press, 35-91, 2015.
PAIVIO,
A. Mental Representations: A Dual Coding Approach. Oxford, Oxford
University Press, 1986.
PICARD,
L., MAYOR-DUBOIS, C., MAEDER, P., KALENZAGA, S., ABRAM, M., DUVAL, C. EUSTACHE,
F., ROULET-PEREZ, E. & PIOLINO, P. Functional independence within the
self-memory system. Cortex, 49, pp. 1463-1481, 2013.
PIOLINO,
P., DESGRANGES, B., & EUSTACHE, F. Episodic autobiographical memories over
the course of time: cognitive, neuropsychological and neuroimaging findings. Neuropsychologia, 47(11), 2314-2329, 2009.
RENOULT,
L., DAVIDSON, P., PALOMBO, D., MOSCOVITCH, M. & LEVINE, B. Personal
semantics: at the crossroads of semantic and episodic memory. Trends in
Cognitive Sciences, 16(11), 550-558, 2012.
Roediger,
H. & McDermott, K. Implicit memory in normal human subjects. In: SPINNLER,
H. & BOLLER, F. (Eds.) Handbook of Neuropsychology. Amsterdam: Elservier,
Vol. 8, 63-131, 1993.
ROSENBAUM,
R., KOHLER, S., SCHACTER, D., MOSCOVITCH, M., WESTMACOTT, R., BLACK, S., GAO,
F. & TULVING, E. The case of K.C.: Contributions of a memory-impaired
person to memory theory. Neuropsychologia, 43,
989-1021, 2005.
ROWLANDS,
M. Memory and the self: Phenomenology, science and autobiography. Oxford
University Press, 2017.
RUBIN,
D. The basic-systems model of episodic memory. Perspectives on Psychological
Science, 1(4), 277-311, 2006.
RUBIN,
D. Event
memory: A theory of memory for laboratory, autobiographical and fictional
events. Psychological Review, 122(1), 1-23, 2015.
RUSSELL,
B. The Problems of Philosophy, 1912. Available at: http://ditext.com/russell/russell.html
SAUNDERS,
J. Does all memory imply factual memory? Analysis,
25(3), 109-115, 1965.
SCHACTER,
D., HARBLUK, J. & McLACHLAN, D. Retrieval without recollection:
An experimental analysis of source amnesia. Journal of Verbal Learning and
Verbal Behavior, 23, 593-611, 1984.
SCHRAUF,
R.W. Bilingual autobiographical memory: Experimental studies and clinical cases.
Culture and Psychology, 6, 387–417, 2000.
Squire,
L.R. Memory systems: A biological concept. In: ROEDIGER, H., DUDAI, Y. &
FITZPATRICK, S. Science of Memory: Concepts. Oxford, Oxford University
Press, 339-344, 2007.
SQUIRE,
L.R., & ZOLA, S.M. Episodic
memory, semantic memory, and amnesia. Hippocampus, 8, 205-211, 1998.
TEKCAN,
A. İ., YILMAZ, E., KAYA KIZILÖZ,
B., KARADÖLLER, D. Z., MUTAFOĞLU, M., & AKTAN ERCIYES, A.
Retrieval and phenomenology of autobiographical memories in blind individuals. Memory,
23(3), 329-339, 2015.
Tulving,
E. Episodic and semantic memory. In: TULVING, E. & DONALDSON, W. (Eds.), Organization
of Memory. New York: Academic Press, 381-402, 1972.
TULVING,
E. Elements of Episodic Memory. Oxford, Oxford University Press, 1983.
TULVING,
E. Memory and consciousness. Canadian Psychology, 26, 1–12, 1985.
Tulving,
E. Organization of memory: Quo vadis? In: GAZZANIGA, M. S. (Ed.). The
Cognitive Neurosciences. Cambridge:
The MIT Press, 753-847, 1995.
Tulving,
E. Study of memory: processes and systems. In: FOSTER, J. K. & JELICIC, M. (Eds.), Memory:
Systems, Process, or Function? New York, NY: Oxford University Press, 11-30,
1999.
TULVING,
E. Episodic memory and common sense: how far apart? Philosophical Transactions of the Royal
Society of London. Series B, Biological Sciences, 356(1413), 1505-1515,
2001.
Tulving,
E. Chronesthesia: awareness of subjective time. In:
STUUS, D. & KNIGHT, R. (Eds.), Principles of frontal lobe functions.
New York, NY: Oxford University Press, 311-325, 2002b.
TULVING,
E. Episodic memory: From mind to brain. Annual Review of Psychology, 53,
1-25, 2002b.
Tulving,
E. Episodic memory and autonoesis: Uniquely human?
In: TERRACE, H. S., & METCALFE, J. (Eds.), The Missing Link in Cognition.
NewYork, NY: Oxford University Press, 4-56, 2005.
VARGHA-KHADEM,
F., GADIAN, D., WATKINS, K., CONNELLY, A., VAN PAESSCHEN, W. & MISHKIN, M.
Differential effects of early hippocampal pathology on episodic and semantic
memory. Science, 277 (5324), 376-80, 1997.
Wheeler,
M.A. Episodic memory and autonoetic awareness. In: TULVING, E. & CRAIK, F.
(Ed.), The Oxford Handbook of Memory. New York, Oxford University Press,
297-312, 2000.
WOLLHEIM,
R. The thread of life. Glasgow, Cambridge University Press, 1984.
YONELINAS,
A. P., & RITCHEY, M. The slow forgetting of emotional episodic memories: an
emotional binding account. Trends
in cognitive sciences,
19(5), 259-267, 2015.