machine nor the machine itself. It is ‘ embodied,’ as the form which
‘ causes matter to be.’ 62
The implications of this belief were well realized by Thomas Aquinas’s
great pupil, Dante, when, at the end of his Paradiso, memory and speech
both fail him. Angels, says Dante, have no need of memory for they have
continuous understanding,63 but human beings must know by remember-
ing physically formed phantasms. Of direct knowledge of God, such as that
indicated at the end of the Paradiso, the memory ‘ can form no adequate
image,’ in Gardner’s phrase (we recall the scholastic aphorism, veritas est
adaequatio rei et verbi), and so it fails the intellect, bringing back nothing but
an inadequate shadow or impression of the vanished vision. 64 The bodily
matrix of memory is also clear in Thomas Aquinas’s belief that sensory
memory does not survive death, for it has ‘‘no activity apart from the
corporal organ.’ 65 The memory that is immortal is what he calls ‘‘intellectual
memory,’ since the intellect does survive death. But this is not true memory.
Intellectual memory is ‘‘the notion of memory’’ (a conception of having had
a memory when one was still alive in one’s body); after death, the soul can
still regard past events which are preserved in its intellect, and it can recall
these (for recollection is an intellectual activity). But it can form no new
memories when it no longer has a body.66 Because memory requires a body,
the souls in Dante’s Inferno are forever stuck in their recollected pasts,
unable to form new memories in hell, yet cut off also from the continuous
vision that nourishes the blessed souls and angels.
Dream-images and memory-images
Other kinds of images were thought to arise from the imagination. It will
be useful to consider these briefly, as a way of distinguishing them from
memory-images. Imagination produces a dream-image in a different man-
ner, though from the same materials. Aristotle says that the mental images
which come in dreams arise spontaneously, not in response to a controlled
process like recollection; in fact this is their chief difference from the
memory-images that are the subject of my study here. Dream-images are
74
The Book of Memory
created by the vis imaginativa, as are all phantasms.67 They are in the same
class as after images, hallucinations, and other irrational images, the
product of aroused, imbalanced emotions (as perception is distorted by
anger or lust) or of raw sense-data unformed by judgment (as when we see
the land moving as we ride through it). Such images are themselves just
sense-data, aisth¯emata, rather than being the imprint of a sense impression
after some time has elapsed, Aristotle’s basic definition of a memory-
phantasm. To put the matter in modern psychological terms, Aristotle
does not consider that a memory is a memory until it has been securely
stored in a retrievable manner in the ‘‘long-term memory.’’ In sleep, the
controlling sensus communis, the seat of awareness, is not functioning. So
the external senses, if stimulated, produce unprocessed raw data; there are
also residual movements, effects, left over from material received during
the day.68 All of this uncontrolled material can stimulate the imagination,
but with the sensus communis and all other judgmental activity in the inner
sense suspended, the resulting mental images pose a special interpretive
problem, for they must be judged after the fact, as it were, as to their truth
or falsity, whether they are of divine origin and predictive of the future, or
whether they are simply the body’s way of adjusting the balance of humors,
or the product of random, raw aisth¯emata.
While Aristotle acknowledges that some dreams can be true, such things
do not seem to interest him very much, and he says little about them. But
Averroe¨s and Avicenna are both greatly interested in prophetic dreams,
which they regard as the province of those with especially well-formed
imaginations. Avicenna discusses the prophetic imagination, ‘ that is, what
is genuinely prophetic in the image-making power.’ 69 Dreams of the future
result from a direct action of angelic intelligences upon the mind in sleep,
acting upon the imagination. He also recognizes inspirations ‘ which sud-
denly fall into the soul [quae subito in animam cadunt],’ unrelated to
previous knowledge or experience. Such inspiration is of various kinds,
‘ sometimes it is from the intellect, sometimes by means of divination, and
sometimes it is from poetry, and this happens according to aptitude and
habit and custom.’ 70 These inspirations, Avicenna says, arise from causes
which assist the soul for the most part unknown and for the most part [they] are
like sudden apparitions, which do not remain in a manner so that they can be
sought out in recollection, unless the mind hastens to aid them with an already-
held memory. 71
Avicenna underlines here the relationship among the images formed in
heightened states such as trances and epilepsy, in dreams, and those
Descriptions of the neuropsychology of memory
75
required in memory. The images produced during dreams and trances will
disappear unless they are associated with images that are already in memory
storage, already familiar and accessible to recollection. Thus even direct
inspiration requires the immediate assistance of human memory, though
in a way more mysterious than that of ordinary dreaming or consciously
controlled recollection.
Meticulous mental imaging – mental painting, really – is a feature of
trained recollection, even when one may be ‘‘recollecting’’ an experience
recorded in someone else’s words. In his Epitome of Aristotle’s Parva
naturalia, written about 1170, Averroe¨s writes of individuals with peculiarly
gifted powers of concentration, who can form accurate images of things
just from their description by others: ‘‘In this manner, it is possible for a
person to form the image of an elephant without his ever having actually
seen one.’’ Such a feat can only be achieved by all three faculties (imagi-
nation, cogitation, and memory) acting together, and is possible only for
humans because it is an intellectual operation. Such concentration is
‘‘exceedingly difficult’’ and ‘‘will only occur in the case of those who
exercise their minds in solitude.’’ Averroe¨s is speaking of consciously con-
trolled procedures here, and it is note-worthy that he believes that the
power of imagining predictively and originally can be achieved by con-
scious meditation as well as by sudden inspirations. But ordinarily it is in
sleep or trance, or in particular cases like epilepsy, that the faculties will all
unite and the dreamer ‘‘will behold the wonders of the world.’’72
Before turning the discussion directly to the processes of recollection, let
me summarize for the sake of emphasis the chief features of a memory-
image. Most importantly, it is ‘‘affective’’ in nature – that is, it is sensorily
derived and emotional
ly charged. It is not simply an abstraction or a
mental ghost, despite its critical usefulness to all rational processes. Nor
does the language of computation adequately describe what a memory was
thought to be – it is not a mere algorithm or schema of the sort that
accounts for what a machine does, though, as we will see, many mnemonic
techniques function, at least in part, like algorithms. But they are never just
that. Successful memory schemes all acknowledge the importance of
tagging material emotionally as well as schematically, making each mem-
ory as much as possible into a personal occasion by imprinting emotional
associations like desire and fear, pleasure or discomfort, or the particular
appearance of the source from which one is memorizing, whether oral
(a teacher) or written (a manuscript page). Successful recollection requires
that one recognize that every kind of mental representation, including those
in memory, is in its composition sensory and emotional. Recollection
76
The Book of Memory
may employ schemes, but it is like reading a book, that is, an event involving
judgment and response (intentio) in addition to intellect. 73
One other feature distinguishes a memory-image from every other sort.
Aristotle says that ‘ Memory is of the past,’’ and ‘‘all memory involves time.’’74
Because time is bound into their nature – memories are presently existing
images of things that are past – memories differ from other sensorily pro-
duced images, as recognition (of something present) differs from recollection
(a matter which I discussed in Chapter 1). Their temporal nature also means
that memory’s re-presentation is less importantly mimetic, or objectively
reiterative of the original perception, than it is temporal, because it makes
the past perception present. Aristotle says that we judge time-lapses and the
relative duration of time by an imaging process similar to that by which we
judge magnitude (449b 30ff., and 452ab), by constructing a sort of scale
model in the mind. Aristotle spends a good deal of his brief treatise on this
particular problem, suggesting a solution which Sorabji describes in pp. 18–21.
What is important to later thinkers about this discussion is Aristotle’s insist-
ence that memorial phantasmata are both representations of things (in the
senses discussed in Chapter 1) and ‘ re-presentations’’ of experience no longer
present. Time is a dimension of all images in the memory.
Thus, recollection was understood to be a re-enactment of experience,
which involves cogitation and judgment, imagination, and emotion.
Averroe¨s and Aristotle both insist on this: ‘‘the one who recollects will
experience the same pleasure or pain in this situation which he would
experience were the thing existing in actuality.’’75 Memory’s success is
heavily dependent on the recollector’s skill in being able to form memory-
images that are ‘ rich’’ in associations, as ‘ iconic’ (to use another term from
neuropsychology) as possible. All mnemonic advice stresses the benefits
to be gained from forming memories as scenes that include personal
associations. Hugh of St. Victor, for instance, stresses the need to impress
the circumstances during which something was memorized as part of the
associational web needed to recall it: the sort of day it is, how one feels, the
gestures and appearance of one’s teacher, the appearance of the manuscript
page, and so on.
R E C O L L E C T I O N
In his Libri rerum memorandarum (1345), Petrarch gives the following
account of the recollective abilities of one of his friends:
It was enough for him to have seen or heard something once, he never forgot; nor
did he recollect only the subject [res], but the words and time and place where he
Descriptions of the neuropsychology of memory
77
had first learned it. Often we spent entire days or long nights in talking: there was
no one I would rather listen to; for after the passage of many years the same things
being spoken of, if I were to say much more or less or say something different, at
once he would gently admonish me to correct this or that word; and when
I wondered and inquired just how he could have known this, he recalled not only
the time in which he would have heard it from me, but in the shade of which holm
oak, by what riverbank, along what sea shore, on the top of what hill (for I had
walked long distances with him along the coasts), I recognizing each particular. 76
Striking in this account is the friend’s mnemonic use of unrelated, expe-
riential detail to mark what we think of as objective information. Some
contemporary psychologists are inclined to think that the memory used in
reproducing information accurately (called ‘‘rote’’ or ‘‘semantic’’ memory)
is unrelated to what is called ‘‘event’’ or ‘‘episodic’’ memory, our ability to
recall events in vivid detail from our own past, because personal memories
are subject to re-creation and inaccuracy, whereas the usefulness of rote
memories (such as remembering that 6 7 ¼ 42) depends exactly on their
unchangingness. But in Petrarch’s account, recollection of personal expe-
riences is used to ensure greater accuracy of memory, an accuracy which he
recognizes when he acknowledges that his friend, recollecting the event as a
whole, was able to correct Petrarch’s own memory of the information –
quotations and learning – they had exchanged. Personalizing bits of
information helps to distinguish one bit from another, and thus is an
aspect of the mental addressing system itself. Each memorized bit, in this
technique, is regarded in the first instance not as information to be
reproduced but as a personal event, with full phenomenological status.
Accuracy comes about through the act of recreating in memory the
complete occasion of which the accurate quotation is a part.
The whole matter of memory error seems to be quite differently con-
ceived by the ancients from the one that fuels modern anxieties about
‘‘making mistakes.’’ For us, making a mistake of memory is a failure in
accuracy, a failure exactly to iterate the original material. In antiquity and
the Middle Ages, problems involving memory-phantasms are described as
heuristic (recollective) rather than as reproductive problems, and are due to
a failure to imprint the phantasm properly in the first instance, thus
causing confusion and recollective loss. One must be careful to form
one’s imagines securely and distinctly in the first place, and by repetition
and practice ensure that they are in long-term memory. One must make
them sufficiently distinct from one another, and use a set order with a
clearly established beginning, like ‘‘one’’ or ‘‘A.’’ One should not tire the
memory by trying to memorize too much at a time, or too quickly, for this
78
The Book of Memory
produces an over-loading problem. Forgetting is a technical error, due to
such things as insufficient imprinting or mis-addressing, and errors of
recollection are thus perceptual in nature, if the min
d’s ‘‘eye’’ cannot see
clearly or looks in the wrong place. But if one’s images are clearly made,
and if one’s routes to them through the mass of individual phantasms
stored in memory are properly marked, and fortified through practice, one
will safely and securely find one’s place. As Quintilian says, ‘‘however large
the number [of items] we must remember, all are linked one to another like
dancers hand in hand, and there can be no mistake since they join what
precedes to what follows, no trouble being required except the preliminary
labor of memorizing’’ (my italics).77 A common image for items associa-
tively grouped in memory is that of catena or chain; perhaps the very
notion of texta itself, which literally means ‘‘something woven,’’ derives
from the same mental phenomenon. And the language that describes the
formation of associations as ‘‘hooking’’ material to other things leads to a
metaphor of recollecting as fishing: as one pulls up one’s line, all the fish on
one’s hooks come with it (as painted in the border shown in figure 28).
The idea also informs the common metaphorical extension in Latin
of the word silva, ‘‘forest,’’ to mean a mass of unrelated and disordered
material. Within his memorial forest, a trained student, like a knowledge-
able huntsman, can readily find the places (loci) where the rabbits and deer
lie. Quintilian observes that:
just as all things do not grow in every country, and you would not find a particular
bird or animal if you did not know its birthplace or its haunts, while even kinds of
fish differ . . . so every Argument is not found everywhere, and we have therefore
to be selective in our search. 78
The spatial nature of this mental search is clearer in the Latin: ‘‘ita non
omne argumentum undique venit ideoque non passim quaerendum est.’’
As the huntsman finds game and the fisherman fish, so the student finds his
stored material – by knowing its habits and habitats. Accuracy is a feature
of the memory-image which is determined at the imprinting stage.
Quintilian counsels having someone read aloud from a text so that one
can check the accuracy of one’s image of it during the process of setting it in
the memory, and to check oneself periodically by re-reading. Evidently
The Book of Memory Page 15