Professional Documents
Culture Documents
Amnesia, Consolidation, and Retrieval: Brooklyn College of The City University of New York
Amnesia, Consolidation, and Retrieval: Brooklyn College of The City University of New York
Amnesia, Consolidation, and Retrieval: Brooklyn College of The City University of New York
ceived information was uniquely encoded in lost. If retrieval is defined as all memory
short-term memory in the form of electro- processes necessary for the utilization of
chemically active, neural reverbatory circuits information already in passive storage, then
that were ultimately terminated by decay a retrieval failure implies that the informa-
and/or intereference processes (e.g., Hebb, tion is present but unavailable on the test
1949). The continued action of these cir- trial. On the basis of these definitions, a
cuits over hours (later minutes or seconds) potential distinction between retrieval and
was believed to give rise to a more perma- consolidation views of experimental amnesia
nent, long-term memory of an energetically is whether or not the memory impairment
passive, chemical-structural format. Elec- is permanent. Some specific models hypoth-
troconvulsive shock and hypothermia were esize further distinctions (e.g., Hebb, 1949;
thought to terminate short-term memory Weiskrantz, 1966), but the issue of perma-
processes before long-term memory could be nence remains primary. The retrieval ex-
established (McGaugh, 1966). Metabolic planation of experimental amnesia would be
inhibitors were believed to block the forma- established by recovering the unavailable
tion of long-term memory during the finite memory in the absence of any subsequent
duration of the relevant information in short- training. On the other hand, the consolida-
term memory (Barondes & Cohen, 1968). tion position could hardly be said to be
In both cases, the consolidation explanation proven by a failure of recovery from amnesia.
of experimental amnesia demands that the The possibility of recovery under untried
induced amnesia be permanent. circumstances always remains. Moreover,
Numerous recent experiments have placed a retrieval failure could theoretically be ir-
in question these explanations and their un- reversible. In the latter case, no behav-
derlying model of memory. For instance, ioral test would suffice to distinguish a re-
information received immediately prior to trieval failure from a consolidation failure,
electroconvulsive shock has been observed and the issue could not be resolved until
to be available for tens of minutes after elec- future technology provided us with tests that
troconvulsive shock (Geller & Jarvik, 1968b; are other than behavioral (e.g., biochemical).
McGaugh & Landfield, 1970; Miller & Clearly, behavioral studies of experimental
Springer, 1971). This suggests that either amnesia are, due to the nature of the para-
electroconvulsive shock fails to destroy all digm, poorly suited for establishing the
information uniquely encoded in short-term validity of consolidation theory.
memory or, alternatively, information enters Most studies have found amnesia to vary
long-term memory far more rapidly than is inversely with the interval between acquisi-
ordinarily assumed. As difficulties with the tion and administration of the amnestic agent
detailed consolidation explanation of the and, given a fixed interval between training
19SO's and early 1960's became apparent, and test trials, to be relatively independent of
many researchers turned to more general the interval between administration of the
models within the consolidation framework amnestic agent and the retention test (after
(e.g., McGaugh, 1968) while others sought correcting for effects of the amnestic agent
explanations of experimental amnesia in dis- on performance variables). These data indi-
ruption of the memory sequence subsequent cate that amnestic agents interfere with time-
to consolidation (e.g., Lewis, 1969). The dependent, memorial processes that are initi-
latter set of explanations is a variation on ated during or soon after acquisition and are
the retrieval failure theme. Few, if any, completed in relatively short order. This
researchers have suggested that retention constraint is consistent with the consolida-
(passive storage) is subject to disruption. tion failure explanation of experimental am-
Consolidation is ordinarily defined as the nesia since all initial entry into long-term
formation of permanent storage of the rele- storage must occur soon after training if
vant information; hence, a prevention of there is to be any retention displayed. How-
consolidation that is prolonged beyond the ever, the constraint is also consistent with a
duration of any temporary memory storage retrieval failure explanation as some retrie-
requires that the information be irrevocably val processes must occur during or soon after
AMNESIA, CONSOLIDATION AND RETRIEVAL 71
acquisition periods to permit later access to poral distinction between consolidation and
the relevant information. As an analogy, retrieval is an operational definition inde-
consider the acquisition of a new book by pendent of any behavioral consequences of
a library. Placing the book on a shelf may the operation, such as the permanence of am-
be viewed as equivalent to consolidation and nesia. Inasmuch as focusing on the perma-
making the appropriate entry in the card nence rather than the temporal distinction
catalog could be considered a retrieval func- between consolidation and retrieval is both
tion, one that must be performed soon after more common and more fruitful, this article
acquisition. Without the appropriate entry will adhere to the former definitions.
in the card catalog, the book may be physi-
cally present in the library but not available RECOVERY FROM AMNESIA
on request. The logical difficulty with consolidation
Returning to memorial processes, the re- explanations of experimental amnesia has
trieval view of experimental amnesia is that always been present. However, the short-
the relevant information is successfully con- comings of the consolidation position were
solidated within the central nervous system far less evident in prior years since there
and that the amnestic agent has impaired were no clear instances on record of recovery
the establishment or future functioning of (of the most recently acquired information)
the cataloging system necessary for retriev- from experimental amnesia. Considerable
ing the information from long-term storage clinical literature reporting spontaneous re-
at some later time, that is, during the reten- covery from traumatic amnesia, oldest mem-
tion test. The retrograde gradient for de- ories returning first, was ignored on the
gree of amnesia as a function of the interval grounds that there was no evidence of re-
between training and administration of the covery for the last few seconds preceding
amnestic agent is, according to the retrieval the traumatic event. The fact that most
position, a measure of the temporally de- memories did return was overshadowed by
creasing vulnerability of some cataloging the less well documented supposition that a
mechanism to the amnestic agent. Clearly few memories did not return spontaneously.
the cataloging mechanism is in itself a mem- The issue of recovery from experimental
ory of some sort; however, it should not be amnesia was first raised by Zinkin and Miller
confused with the fully elaborated memory (1967) who observed partial spontaneous
of the training event, which may exist inde- recovery from electroconvulsive shock-in-
pendently of catalog-type memories. Cata- duced amnesia when retention intervals
log-type memories are assumed not to be greater than the traditional 24 hours were
sources of primary information, but rather used in a repeated-tests paradigm. Subse-
means to facilitate access to primary in- quent studies have frequently failed to repli-
formation. One hypothesis to explain why cate this phenomenon, particularly when the
amnestic agents act preferentially on cata- possibly confounding repeated-tests aspect
loging processes rather than on consolidation of the experiment was removed (Herz &
is to assume that consolidation of primary Peeke, 1968; Luttges & McGaugh, 1967).
information occurs extremely rapidly, per- While spontaneous recovery from electro-
haps in fractions of a second, while catalog- convulsive shock-induced amnesia seems a
ing occurs over seconds, minutes, or even doubtful phenomenon, several recent studies
hours. Data and arguments supporting this have reported extensive spontaneous re-
hypothesis comprise much of the remainder covery from amnesia induced by the protein
of this article. synthesis inhibitor cycloheximide (e.g.,
As an alternative to the above schema, a Quartermain & McEwen, 1970).
few investigators have defined all events Koppenaal, Jagoda, and Cruce (1967)
occurring during acquisition as uniquely part initiated a second approach to the problem
of the consolidation process and all events of the permanence of electroconvulsive
occurring during testing as retrieval (Cher- shock-induced experimental amnesia. After
kin, 1970; Weiskrantz, 1966). While one training and amnesia-producing treatment,
can not dispute a definition, such a tem- they subjected their animals to a second
72 RALPH R. MILLER AND ALAN D. SPRINGER
training trial (reminder) with reinforcement with electroconvulsive shock can be largely
intensity sufficiently reduced from initial restored with scopolamine injected just prior
training such that animals receiving only to testing. Unfortunately, the permanence
the second training trial learned little or of the restored memory in this study was
nothing. The amnestic animals, however, not examined; however, even a temporary
displayed considerable recovery of memory. return of memory indicates that the informa-
Geller and Jarvik (1968a) successfully repli- tion was previously present although un-
cated this phenomenon. The difficulty with available. Another example is provided by
both of these examples of recovery is that Flexner and Flexner (1967) who, under
the initial training and amnesia-producing select circumstances, produced significant re-
treatment may have created an animal devoid covery from puromycin-induced amnesia in
of the relevant memory but with a superior mice by intracerebral injection of saline.
learning potential during the second training The explanations of these studies are far
trial. Lewis, Misanin, and Miller (1968) from clear. Davis, Thomas, and Adams
observed a significant return of memory (1971) theorize about the importance of
after correcting for this ambiguity by adding appropriate cholinergic levels for retrieval
control groups and administering the re- paralleling Deutsch's (1971) cholinergic
minder stimuli under noncontingent circum- model of spontaneous forgetting. However,
stances (outside of the training-test environ- all such recovery studies add weight to the
ment) that prohibited learning during the retrieval view of experimental amnesia which
reminder trial. Miller and Springer (1972a) postulates that the relevant information is
replicated this phenomenon and further present but unavailable.
demonstrated that the apparent recovery of The stimulus used to induce recovery of
memory was truly memorial in nature rather memory in most studies of facilitated re-
than a systemically induced phenomenon covery from electroconvulsive shock-induced
such as an alteration in activity level. amnesia has been the same aversive rein-
Quartermain, McEwen, and Azmitia forcer as was used in training. The question
(1970) reported studies corroborating the of stimulus specificity of the recovery agent
effectiveness of a noncontingent presentation with respect to the training reinforcer was
of the training reinforcer in restoring mem- recently raised by Springer and Miller
ory after either cycloheximide or electro- (1972). They used a one-trial passive
convulsive shock. They produced consider- avoidance task with ice water immersion as
able recovery of memory only when the the reinforcer. Rats receiving electrocon-
reinforcer (footshock) was presented within vulsive shock immediately after training dis-
an hour after an initial retention test. How- played essentially complete amnesia, but
ever, Miller and Springer (1972a) observed appreciable recovery of memory was ob-
extensive recovery even when a retention served if they were subjected to a noncon-
test was not given to the recovery group tingent footshock between electroconvulsive
prior to presentation of the noncontingent shock and the test trial. Control animals
reminder. The same study also observed indicated that the observed avoidance was
significant recovery from anmnesia over not due to changes in activity levels. Non-
electroconvulsive shock-reminder intervals of specificity of the recovery agent for electro-
up to 2 weeks (the longest interval ex- convulsive shock-induced amnesia is also
amined). Consequently, we are inclined to evident in the previously mentioned scopola-
reject the notion of an absolute critical period mine-induced recovery of memory reported
for recovery-inducing agents. by Davis, Thomas, and Adams (1971) and
Numerous other studies have found a di- is consistent with the prevention of cyclo-
verse group of stimuli and pharmaceutical heximide-induced amnesia achieved with
agents that induce partial or total recovery amphetamine and corticosteroids by Bar-
from experimental amnesia. Typical of the ondes and Cohen (1968), although in the
pharmacological studies are the experiments latter case these stimulants were effective
of Davis, Thomas, and Adams (1971) that only when injected soon after training and
indicate memory in rats rendered amnestic cycloheximide treatment.
AMNESIA, CONSOLIDATION AND RETRIEVAL 73
The results of the Springer and Miller propriately long exposure to the training
(1972) study suggested that the word re- environment as an effective reminder.
minder may be a misnomer for many stimuli The multitrial training used by Adams
that can induce recovery from amnesia. (1966) differs significantly from the one-
What this study did not make clear is trial training tasks that dominate the more
whether any stressor can act as a recovery recent literature. Consequently, to examine
agent regardless of the training task. Or Adams' effect in a contemporary context,
alternatively, is a stimulus similarity of the Miller, Springer, and Vega (1972) gave
internal physiological state produced by the one-trial aversive training in a shuttle box
typically aversive training reinforcer and the followed by immediate electroconvulsive
stressful recovery agent, thus permitting the shock and found that memory could be
latter state to act as a generalized reminder ? largely restored with brief exposure to the
To resolve this issue, Miller, Springer, and punishment compartment or somewhat longer
Vega (1972) used a one-trial appetitive task exposure to the start compartment. How-
and administered electroconvulsive shock im- ever, the data suggested that overexposure
mediately after training. Several parametric to either compartment not only did not
studies varying noncontingent recovery further improve retention but could actually
shock level failed to restore memory con- impair retention. Seemingly, after memory
trary to the high effectiveness of noncon- has been restored, it is vulnerable to extinc-
tingent shock in restoring memory of aver- tion like any normal memory. The potential
sive events. This suggested that the stress of apparatus cues to restore memories after
of recovery shock in Springer and Miller administration of an amnestic agent suggests
(1972) was specifically acting as a reminder that memory could reappear over repetitive
of earlier stressful events rather than as a test trials. In this light, the Zinkin and
reverser of all potential amnesias. Further Miller (1967) study, frequently ignored
experiments by Miller, Springer, and Vega exactly because of its multiple test trials,
(1972) found that appetitive memories could becomes worthy of reexamination. Perhaps
be largely restored after electroconvulsive these data are not an example of spontaneous
shock through either brief (1 minute) non- recovery but rather recovery facilitated by
contingent presentation of the appetitive re- intervening test trials that serve as re-
inforcer or somewhat longer exposure (three minders. Moreover, the shade of doubt
exposures of 1 minute each) to the training surrounding these data due to the failure
apparatus in the absence of the reinforcer. of others to replicate them (e.g., Luttges &
Looking for apparatus cues that could McGaugh, 1967) is eased by the recent suc-
restore aversive memories after electrocon- cess of Quartermain, McEwen, and Azmitia
vulsive shock brought to light a study by (1972) in restoring considerable memory
Adams (1966) in which amnesia for mul- with repeated test trials after either cyclo-
tiple-trial, aversive training in a shuttle box heximide or electroconvulsive shock treat-
was greatly attenuated through three ex- ment.
posures of 5 minutes each to the start com- Collectively, these recovery of memory
partment. Adams used a classical condition- studies indicate that at least some instances
ing explanation of electroconvulsive shock- of experimental amnesia are cases of re-
induced amnesia (US = electroconvulsive trieval failure and that many of the recovery
shock, UR — convulsion, CS = apparatus agents may reasonably be called reminder
cues, CR = partial covert convulsion) and stimuli. Until there is concrete evidence for
explained his data as an instance of extinc- two different sources of experimental am-
tion. Due to other data, the classical con- nesia, it would seem reasonable to assume
ditioning position has been discounted by the generality of the retrieval failure hypoth-
most researchers today (e.g., McGaugh & esis. However, an alternative to the unipro-
Petrinovich, 1966) ; however, consolidation cess retrieval hypothesis has been proposed.
theory offers no alternative explanation of The hypothetical argument has frequently
Adams' data. The retrieval position can been made that if the amnestic agent (e.g.,
interpret the phenomenon by regarding ap- electroconvulsive shock) were "effectively
74 RALPH R. MILLER AND ALAN D. SPRINGER
appetitive and aversive reinforcers (e.g., and potentiate the memory loss caused by
Davis & Hirtzel, 1970; Davis & Klinger, an amnestic agent administered immediately
1969; DeVietti & Larson, 1971; Howard after detention. To cite a few examples,
& Meyer, 1971; Lewis, Bregman, & Mahan, Rbbustelli and Jarvik (1968) found that
1972; Misanin, Miller, & Lewis, 1968; detention caused amnesia in mice and, if
Schneider & Sherman, 1968). There have followed immediately by electroconvulsive
also been failures to obtain the reactivation shock, potentiated the amnestic effect of
phenomenon (Banker, Hunt, & Pagano, electroconvulsive shock. Davis and Klinger
1969; Dawson & McGaugh, 1969; Gold & (1969) and Davis and Hirtzel (1970), using
King, 1972; Jamieson & Albert, 1970); goldfish, observed that detention potentiated
however, none of the studies reporting null the amnesia produced by KC1, puromycin,
results distinguish themselves by making acetoxycycloheximide, and electroconvulsive
extensive parametric attempts to find the shock. And Geller, Robustelli, and Jarvik
optimal conditions for the phenomenon. (1971) found that posttraining detention
Clearly all of the variables controlling the of mice produces a memory deficit that is
phenomenon are not understood; however, additive to one produced by cycloheximide.
the pervasiveness of the effect across tasks, Considering detention as a performance
species, amnestic agents, and independent attenuating treatment in its own right, the
laboratories argues for the validity of the question arises as to whether detention is
phenomenon. merely an example of extinction (cessation
Considering the long training-electrocon- of previous reinforcement contingencies as-
vulsive shock interval used in reactivation sociated with training cues) and/or retro-
studies, consolidation failure appears to be active interference (due to irrelevant in-
a most unlikely explanation of the observed formation input) or whether it is a new
memory deficit. However, the administra- phenomenon in and of itself. Two variables
tion of an amnestic agent in conjunction that facilitate understanding of the detention
with select stimulus configurations at any effect are the place of detention and the delay
time prior to behavioral testing could, in between training and detention. Robustelli
principle, prevent the normal functioning of and Jarvik (1968) presented data suggesting
retrieval processes during the test trial. The that detention becomes less effective as an
occurrence of experimental amnesia pro- amnestic agent as the stimuli during deten-
duced at the time of reactivation further tion become increasingly divergent from
suggests that at least some experimental those of the training apparatus. This is
amnesia is due to a retrieval failure, a con- consistent with detention effects resulting
clusion in concurrence with the studies re- from extinction (to those conditioned stimu-
porting recovery from amnesia. Again, if lus elements present during detention) and
some experimental amnesia is due to re- retroactive interference (provided interfer-
trieval failure, parsimony compels us to ac- ence is assumed greatest when the irrelevant
cept all experimental amnesia as resulting information concerns aspects of the context-
from retrieval failure until evidence to the ual cues of the training situation). Calhoun
contrary is presented. Such evidence, due and Murphy (1966) presented data support-
to experimental design, cannot be provided ing this assumption for a passive avoidance
by any experimental amnesia paradigm used task similar to that used by Robustelli and
to date. Jarvik (1968). Similarly, the memory deficit
produced by detention increases with dimin-
DETENTION
ishing intervals between training and deten-
Despite apparent discrepancies in the para- tion (e.g., Robustelli & Jarvik, 1968). As
digms used, the recent studies of posttraining extinction and interference are most effective
detention are closely related to both re- soon after acquisition (Skinner, 1938), this
covery and reactivation phenomena. Several too is congruent with detention being re-
research groups have recently reported that garded as a combination of these two factors.
detaining an animal in the training apparatus Moreover, a spontaneous decrement in de-
after acquisition can both produce amnesia tention-induced deficits over a 24-hour, post-
76 RALPH R. MILLER AND ALAN D. SPRINGER
returning the subject to the altered state as the time of consolidation, the only serious
shortly before a retention test has been seen question can be what are the temporal pa-
to restore the previously unavailable memory rameters ?
(Thompson & Neely, 1970). However, the In addition to interfering with retrieval,
restoration data supporting this seemingly electroconvulsive shock may reasonably be
successful model have been generated by a assumed to produce neural-electrical silence
paradigm similar to the one used in the pre- during the subsequent comatose state, or, at
viously described reminder studies with the least, to inject sufficient noise into the neural
stimuli affecting the return to the altered system to obliterate any information encoded
state serving as the recovery agent. The uniquely in electrochemical transmission
basic distinction between this state-depen- (unconsolidated memory). Therefore all in-
dent model and the reminder model is formation surviving electroconvulsive shock,
that the state-dependent position predicts whether immediately available or observable
that if the retention test is delayed a day or only after administration of a recovery agent,
more after the animal has been returned to must have already achieved a structural-
the altered physiological state for the second chemical format. In view of the recovery
time, memory should once again be unavail- data, this suggests extremely short consoli-
able. Miller, Malinowski, Puk, and Springer dation times. In fact, Lewis, Miller, and
(1972) examined this prediction and found Misanin (1969) report that under select
that memories once recovered do not fade circumstances (noncontingent familiarization
again over several days. Moreover, Miller, with the training environment prior to train-
Springer, and Vega (1972) found some ing) even a 500-millisecond interval between
effective recovery agents that do not subject training onset and electroconvulsive shock
the amnestic animal to the drastic change onset failed to yield amnesia. Lack of am-
in physiological state that scopolamine, foot- nesia following pretraining environmental
shock, and/or electroconvulsive shock pro- familiarization has also been obtained with
duce. Consequently, this second version of hypothermia as the potential amnestic agent
the state-dependent hypothesis is also found (Jensen & Riccio, 1970). These data argue
to be inadequate. Experimental amnesia for the occurrence of consolidation in inter-
appears to reflect a retrieval failure, but does vals of less than 500 milliseconds under these
not seem to be a dissociated learning phe- select circumstances and perhaps more gen-
nomenon. erally. Such consolidation intervals may
seem extremely short by traditional stan-
CONCLUSION dards, but numerous metabolic reactions in
The preceding arguments in support of a the nervous system are completed in equiva-
retrieval explanation of experimental am- lent temporal spans.
nesia are not meant to refute the concept of One phenomenon not readily understood
consolidation but rather to question its appli- in either a consolidation or a retrieval frame-
cability to the amnesia data. Assuredly, in- work is the short post-electroconvulsive
formation impinging on an organism is initi- shock interval during which memories ac-
ally represented by electrochemical, neural quired immediately prior to electroconvulsive
transmission. At much later periods of time, shock are available (e.g., McGaugh & Land-
this information is apparently stored in some field, 1970; Miller & Springer, 1971). Both
energetically passive, structural-chemical for- the consolidation and retrieval positions
mat (as, for example, evidenced by retention would be satisfied by hypothesizing that elec-
of all but the most recently acquired informa- troconvulsive shock either prevents consoli-
tion over a period of extreme hypothermia dation or future retrieval but does not de-
and consequent neural-electrical silence stroy short-term memory that is seen to be
(Lipp, 1964; Riccio, Hodges, & Randall, increasingly attenuated over time due to
1968)). Ignoring the possibility of more either decay or interference. This possi-
than two states of memory (short-term mem- bility, however, seems most unlikely if short-
ory and long-term memory), if the initial term memory is encoded in electrochemical
establishment of the latter state is defined neural transmission, for such a format would
78 RALPH R. MILLER AND ALAN D. SPRINGER