did not. In Money cooperation was based on monetary trade, which is self- enforcing by design. A producer has an incentive to help for a token—to avoid being “liquidity constrained” in the future—and has nothing to gain from helping free-riders, who have nothing to offer in exchange. This contrasts with the record-keeping system in Memory because opportunistic consumers can always “pay” by accumulating negative balances. Here, cooperation is self-enforcing if no producer helps such opportunistic individuals, yet there is a temptation to do so because the producer “gets paid” after all. A failure to punish is the source of a negative externality, which magnifies free riders’ opportunistic motivations and displaces cooperation. In Memory, subjects failed to fully appreciate such externality and often failed to punish. Instead, token exchange in Money internalized this externality. As a result—although liquidity constraints are generally considered a source of inefficiency—relaxing them in the experiment (as we do in Memory) lowers long-run efficiency (Result 3). In our laboratory economies liquidity constraints impose discipline on sanctioning behavior, which channels the group toward cooperation.
We analyzed data from three study samples: (a) insomnia sample from study 1, (b) insomnia sample from study 2, and (c) healthy sample from study 2. All participants performed word-pair (A –B) declarative memory task in the evening (evening retrieval) before as well as in the morning (morn- ing retrieval) after 8 hr of nocturnal sleep. Based on their performance during the evening retrieval session, we calcu- lated their individual trait measure of BMP. Each of the three study samples was divided with median split into two BMP groups: moderate (BMP −) and high (BMP+) perfor- mers. It is to note that the design of the two studies included in our analyses slightly differed. Especially, while all the participants from the study 1 underwent in the evening the same amount of learning and retrieval session, participants from the study 2 were retrieving word pairs until they reach the criterion of 70% correctly recalled material. Moreover, in the two studies, participants were requested to encode different amounts of the word pairs: 80 in the study 1 and 60 in the study 2. In this paper, we investigated differences in memory retrieval in BMP groups with respect to changes in sleep architecture, SpA, and SpD as well as density and peak-to-peak amplitude of SO.
By cooling one part of cortex while recording from another in the course of memory tasks, we were able to obtain the first evidence of tonic in fluences from one cortical region upon another. These tonic influences seem to be part of the mechanism for the maintenance of short-term memory in wide cortical networks (Fuster et al., 1985). The monkeys for our experiment were trained to perform a visual memory task: delayed matching to sample with colors. On every trial, the animal had to retain a color in memory for some 1 0 - 2 0 sec (delay) in order to perform a correct color match at the end of the trial. We knew from previous experiments (Fuster and Jervey, 1982; Fuster et al. 1982) that the memorization of the color stimulus increases in a sustained manner the activity of large populations of cells in the infero temporal and prefrontal cortices. Some of the cells, in either cortex, show a different level of activa tion depending on the color in memory. We also knew that the cooling of either of those cortices, but not parietal cortex, induced a reversible deficit in the performance of the task (Bauer and Fuster, 1976; Fuster et al., 1981).
and adults (20–30 years) in a declarative task, where they had to learn new associations between non-objects and their func- tions. Retrieval performance was tested during an immediate and a delayed retrieval session separated by either a retention interval containing sleep or wakefulness. Sleep led to stabi- lized memory retrieval performance only in children, not in adults, whereas no age-related difference was observed after a similar period of wakefulness. The authors concluded that this effect might be related to more abundant and deeper sleep during childhood. Another very interesting behavioral study by Prehn-Kirstensen and colleagues [ 14 ] showed that sleep prevents the forgetting of reward-associated memory repre- sentations. They found an indication that this effect is more pronounced in children than in adults. It has to be noted— like for other studies investigating sleep-related memory in children—that besides these very interesting behavioral ef- fects, the authors of both publications did not record polysomnography during sleep and therefore could not quantify sleep on a physiological level, or search for poten- tial relationships with overnight changes in memory perfor- mance. However, studies like these, comparing populations of adults with that of children, are rare but are critical to understanding whether and how these processes differ in adults as compared to children. In addition to studies reporting solely behavioral results, there is second category of empirical studies, which apply polysomnography in chil- dren. These studies mainly report results of a purely correl- ative nature, where sleep parameters are correlated with learning and memory performance to uncover the underly- ing neurophysiological correlates of the beneficial effect of sleep for memory consolidation. One very recent study [ 15 ••] addressed the association of sleep electroencepha-
In the Money treatment, monetary trade is self-enforcing because of the presence of liquidity constraints. A producer has an incentive to help for a token—to avoid running out of tokens in the future—and has nothing to gain from helping free-riders, who have nothing to offer in exchange. On the other hand, in Memory and Money Unconstrained opportunistic con- sumers can always “pay” by accumulating negative balances. Cooperation would be self-enforcing if no one helped opportunistic individuals, but the in- centive to do so is too weak because producers who help “get paid” in any case. It is precisely this lack of punishment that generates a negative ex- ternality, which magnifies free riders’ opportunistic motivations and displaces cooperation. As a result—although liquidity constraints are generally consid- ered a source of inefficiency—relaxing them in the experiment (as we do in Memory and Money Unconstrained) lowers long-run efficiency (Results 3 and 7). In our laboratory economies liquidity constraints impose discipline on sanctioning behavior, which channels the group toward cooperation.
recalled colorful and moving images in memory of yesterday. They still recalled such images when they retrieved impressive events in remote memory. However, they also recalled some stationary and black & white images in the retrieval of the remote past. In contrast, patients with AD did not recall colorful and moving images as the controls did in the retrieval of memory of yesterday. They also had great difficulty recalling mundane daily life events. This deficit also happened to a patient in the earliest stage of AD who can still attend hospital on his/her own as described in Chapter III. Impairment in arranging past events in order of occurrence has been discovered by several previous studies (Johnson & Kesner, 1997; Storandt et al., 1998; Lawlor et al., 2004). The present study also revealed that the patients in the early stage of AD had impaired time order memory. In Chapter III, a subject with mild depression was shown to recall fewer images compared to healthy elderly subjects, but reproduced exact details of each image. Their problem of memory retrieval might be more related to low attention resulting from cynicism, apathy and irritability than the memory loss that can be observed in patients with AD. Bergouignan et al. (2006) suggested that episodic autobiographical memory is not impaired in patients with remitted depression.
oral – group interaction, and written history, which comprises any accounts of the past that are laid down and fixed with the aid of storage media. Halbwachs insisted that the notion of memory should only pertain to personal experiences which are reconstructed from the group’s point of view, supple- mented by memories that are narrated by group members with which one interacts, typically parents and grand-parents. History, by contrast, is based on non-personal, materialized forms of storage that enable the invariant presentation of the same account of the past to a diffuse public. Even the most popular media productions – Halbwachs gave films and theatre pieces on Jeanne d’Arc as examples – remain abstract and foreign to collective memory, because they do not connect to lived experience, because “I cannot go beyond these word heard or read by me”, because “these symbols passed down through time are all that comes to me from that past” (Halbwachs 1980: 52). Whereas memory presupposes the continuity of past and present, history marks a rupture. This point is stretched to the extreme when the simple fact of putting the past into writing is taken to indicate that it has lost its value for group memories: “General history starts only when tradition ends and the social memory is fading or breaking up. So long as a remembrance continues to exist, it is useless to set it down in writing or otherwise fix it in memory.” (Halbwachs 1980: 78) Accordingly the society addressed by historiography will never be the group that experienced the events in the past.
This paper presents a dynamical-systems extension of time-discrete sequence memory models with inhibitory feedback. The effect of instantaneous global feedback inhibition in memory networks has been well studied (e.g. Tsodyks and Feigel’man 1988 ; Golomb et al. 1990 ; Treves 1990 ; Hirase and Recce 1996 ; Amit and Huang 2010 ). Our model shows that also dynamical feedback inhibition can stabilize the retrieval of memory se- quences and thereby increase both memory capacity and robustness. The optimal instantaneous global in- hibitory feedback is a roughly linear function of the total network activity as numerically found by Hirase and Recce ( 1996 ) and semi-analytically confirmed by a probabilistic approach in the present paper. Extending the model to dynamic global inhibition, we find that, at the edges of stable replay, inhibition induces strong oscil- lations, which can be interpreted as gamma oscillations. Gamma oscillations are ubiquitous in the brain and their origin is generally attributed to local inhibitory networks (Wang 2010 ). Several cognitive functions have been related to increased gamma power and co- herence, such as sensory integration, attention, and memory (Jutras and Buffalo 2010 ). Specifically, gamma coherence between subregions in the hippocampal for- mation and prefrontal cortex has been shown to cor- relate with involvement in a short-term memory task (Sigurdsson et al. 2010 ). This finding fits well into the general view of gamma rhythms as a mechanism that facilitates communication between brain areas (e.g. Colgin 2011 ). In our model, gamma occurs as a side effect of feedback stabilization during replay. In com- bination with these findings, our model suggests that memory networks may have to be critically loaded to be able to transfer information to other brain areas.
based on some memory mechanism. On shorter time scales, the impression of object constancy, e.g., during a temporary occlusion, provides an exam ple of another memory function that starts to ap pear in the human infant only a few months after birth. On even shorter time scales, the perception of continuous motion and our ability to predict its imminent course can be viewed as a form of a short time memory specialized for this type of task. Finally, all kinds of learning processes, rang ing from the formation of simple conditioned re flexes over the chaining of elementary motor pro grams to the elaboration of complex systems of associatively interrelated concepts are all based in an essential way on some type of memory mecha nism.
Some marched through the streets carrying banners that read “Bombing Holocaust.” And some set up a series of large posters bearing the names of the cities Dresden, Nagasaki, New York, and Baghdad. In these com- memorative acts, one event, the bombing of Dresden, was associated with three completely different messages: one diplomatically conciliatory, one aggressive and vengeful, and one pacifistic. Memories are constantly as- sociated with arguments, but the arguments are never an intrinsic part of those memories. To neutralize the malignant potential of memories, a line must be drawn between what has been experienced and what follows from the experience in terms of interpretation, evaluation, claim, and consequence. The same holds true of the assessment of the year 1945 in German memory. After decades of considering it in terms of “catastro- phe” and “downfall,” the notion of “liberation” was introduced and took hold in the heads of the younger generations. Again, it is not the events that we have changed, but our frames for interpreting them.
Many-body classical memories based on dissipation (often under the name of asynchronous celular automata) have naturally appeared in the context of classical fault tolerant computa- tion. For example, using a simple local update rule on a 2D lattice, Toom [119, 50] showed that classical information can be protected against weak local noise. A more elaborate update rule by G´ acs  provide protection even on a 1D lattice. These results already suggest that dissipation may offer a powerful alternative to the existing methods for constructing many- body quantum memories, as investigated in the present work. In fact, several authors have already proposed the use of continuous quantum error correcting codes [105, 5, 109, 101, 87]. However previous works concentrate on a single level of error correction and do not address the large N many-body scenario. A notable exception is the work of Dennis et al.  intro- ducing a heat bath algorithm (thermal dissipation for the 4D toric code) in order to simplify the efficacy analysis of a local many body quantum error correction algorithm. At the crux of this approach is that thermal dissipation can be interpreted not only as introducing deco- herence (errors), but also as performing a form of error correction, with the balance between the two effects roughly given by the bath temperature. Indeed, this heat bath algorithm can already be seen as a dissipative quantum memory lending itself to more natural engineering. In fact, engineered dissipation is more general in that it need not satisfy detailed balance conditions and thus its power extends that of cooling a self–correcting Hamiltonian. In other words, the steady state need not be an equilibrium state and its dynamics may show a net flow (imagine a funnel receiving water from a hose). As the classical results show, this more general kind of dissipation may be crucial in order to correct general kind of errors.
Commemorating the past cannot be done in a forced manner; it should not represent a voluntary revival of a dying ideology, nor should it be done as a sterile act to clear the country’s conscience. It should render the events neither sacred nor banal. It should not succumb to nostalgia nor water down what happened in the sea of time and forgetfulness into which so many other events have disap- peared. The UNAM’s memorial attempts to confront its spectators with the peculiar power of memory while retaining the necessary distance from the dullness of predetermined value judgments. Aimed primarily at human nature, including its contradictions, it construes this power of memory as a creative experience.
As a result, the more disorder there is in the system, the weaker are the memory effects. From this interpretation, the non-Markovity parameter acquires the quality of a disorder measure comparable with configuration entropy. The direct relation between the non-Markovity parameter and entropy can be received from the scaling relationship between the diffusion coefficient and the excess entropy S = (S liquid − S id.gas )/ k B suggested by Dzugutov [ 30 ]. In the original formulation
conditioned and/or conscious behavior is called for in order to understand how memory can persist over a longer period of time. Eric R. Kandel has, again, been instrumental in clarifying the neuro- physiological mechanism responsible for „long term memory“. Surprisingly, it hinges on an anatomical change on part of the sensory neurons. While a single stimulus strengthens the synapse involved, repeated stimulation triggers a further chemical process within the cell. It is directed towards the genetic information residing in the cell's nucleus and results, ultimately, in the growth of additional synapses. Long term facilitation of biological senso-motor linkage is, as it turns out, supported by structural changes in the cell's architecture.
Cinephilia take one, I suggested, is a discourse braided around love, in all the richly self-contradictory, narcissistic, altruistic, communicative and autistic forms that this emotion or state of mind afflicts us with. Film studies, built on this cinephilia, proceeded to deconstruct it, by taking apart mainly two of its key components: we politicized pleasure, and we psychoanalyzed desire. An important task at the time, maybe, but not a recipe for happiness. Is it possible to once more become innocent and political? Or to reconstruct what, after all, cinephilia take one and take two have in common, while nonetheless marking their differences? The term with which I would attempt to heal the rift is thus neither pleasure nor desire, but memory, even if it is no less contentious than either of the other two. At the forefront of cinephilia, of whatever form, I want to argue, is a crisis of memory: filmic memory in the first instance, but our very idea of memory in the modern sense, as recall mediated by technologies of re- cording, storage, and retrieval. The impossibility of experience in the present, and the need to always be conscious of several temporalities, which I claimed is fundamental to cinephilia, has become a generalized cultural condition. In our mobility, we are “tour”-ists of life; we use the camcorder with our hands or often merely in our heads, to reassure ourselves that this is “me, now, here.” Our experience of the present is always already (media) memory, and this mem- ory represents the recaptured attempt at self-presence: possessing the experi- ence in order to possess the memory, in order to possess the self. It gives the cinephile take two a new role – maybe even a new cultural status – as collector and archivist, not so much of our fleeting cinema experiences as of our no less fleeting self-experiences.
Digitization thus radically alters the storage sites of cultural memory, in both temporal and material terms. While phonographic cylinders and vinyl re- cords house dynamic cultural memories that require motion to be activated, they are themselves relatively stable modes of inscription: apart from the en- tropy that befalls any earthly material, they remain in a stable material state. The digital, as Wendy Chun notes in her astute analyses of the storage capa- bilities of digital archives, is different: “If our machines’ memories are more permanent, if they enable a permanence that we seem to lack, it is because they are constantly refreshed – rewritten – so that their ephemerality endures, so that they may ‘store’ the programs that seem to drive them” (Chun 2011: 170). Digitization thus changes the nature of sound storage from a relatively stable mode of preservation to highly dynamic processes of constant regeneration. Cultural memories are no longer preserved in static terms, but rather depend on constant infrastructural migration. Hence, electronic memories represent a paradox: they become more permanent the more they are constantly refreshed, so that only “their ephemerality endures.” (ibid.)
estimates of beta, as also acknowledged by the authors themselves. Deo & Hurvich
( 2001 ) and Arteche ( 2004 ) show that for perturbed series any inference on the order of integration is biased such that the series appear to be less integrated. Our beta estimates based on intra-day observations, on the other hand, are less noisy, implying that the true order of integration can be better detected. To further illustrate this, one might think of comparing the 2ELW estimates to estimates made by noise robust estimators such as those of Sun & Phillips ( 2003 ) or Frederiksen et al. ( 2012 ). However, these are positively biased when the sample size is smaller than 500, making them inappropriate for our setup. As an alternative we show in Table A.2 of the Online Appendix that changing the bandwidth m in the 2ELW estimation leads to similar estimates of d. As demonstrated by Hurvich et al. ( 2005 ), this would not be the case if the series were seriously perturbed. Second, Andersen et al. ( 2006 ) rely on graphical investigation of the first 36 autocor- relations instead of consistent estimation of the memory parameter. Particularly in small samples ( Andersen et al. , 2006 consider T = 148) this type of inference may lead to false conclusions. We illustrate this by means of a small simulation study for which we report the results in Table A.1 of the Online Appendix. We simulate fractionally integrated noise, i.e., (1 − B) d y t = t , with B being the backshift operator, for memory parameters
This paper experimentally studies the role of imperfect and associative memory for the formation of expectations or beliefs. In the textbook model of Bayesian updating, mem- ory imperfections play no role: agents entertain a prior belief, update this belief upon receipt of information, and yesterday’s posterior equals today’s prior. Our paper starts from the premise that people do not constantly entertain beliefs about every potentially relevant state of the world. Rather, when people are prodded to act on or update their beliefs, they need to reconstruct their prior knowledge and beliefs from memory. This observation raises the empirical question how people retrieve prior information, and which features of news make it more or less likely for memory traces to get recollected. The second observation that motivates our paper is that real-world information sig- nals typically do not just consist of abstract information. Rather, information is often embedded in memorable contexts, including stories, images, emotions, or sounds. More- over, similar news are frequently embedded in similar contexts. For example, when individ- uals receive negative feedback about their performance, these negative news are often associated with scolding, public shaming, and emotions of insufficiency. Similarly, when good news prevail in the stock market, people are disproportionately exposed to bulls, upward-sloping trend lines, and good-times stories.¹ To take yet another example, when immigration opponents relay negative information about the “typical” character traits of immigrants, then this often occurs through similar stories and images involving theft and other forms of violence.
Gourieroux and Jasiak ( 2001 ) show that breaks in the mean of a time series can also lead to biased estimates of the covariance structure of the process again falsely indicating long memory.
Other than the before mentioned papers K¨ unsch ( 1986 ), Parke ( 1999 ) and Diebold and Inoue ( 2001 ) construct diverse processes which do share the autocorrelation structure with a long-memory time series at least in finite samples. K¨ unsch ( 1986 ) considers monotonic deterministic trends. Parke ( 1999 ) constructs a random level shift process where the shift probability is derived from a heavy tailed distribution. A similar process is considered in Mikosch et al. ( 2002 ). This process has the same autocorrelation function as a long memory process. However, Davidson and Sibbertsen ( 2005 ) show that its partial sums converge to a Levy motion with independent increments not exhibiting long memory. Only cross-sectional aggregation of these processes generates long memory, asymptotically.