Auditory short-term memory behaves like visual short-term memory

Are the information processing steps that support short-term sensory memory common to all the senses? Systematic, psychophysical comparison requires identical experimental paradigms and comparable stimuli, which can be challenging to obtain across modalities. Participants performed a recognition memory task with auditory and visual stimuli that were comparable in complexity and in their neural representations at early stages of cortical processing. The visual stimuli were static and moving Gaussian-windowed, oriented, sinusoidal gratings (Gabor patches); the auditory stimuli were broadband sounds whose frequency content varied sinusoidally over time (moving ripples). Parallel effects on recognition memory were seen for number of items to be remembered, retention interval, and serial position. Further, regardless of modality, predicting an item's recognizability requires taking account of (1) the probe's similarity to the remembered list items (summed similarity), and (2) the similarity between the items in memory (inter-item homogeneity). A model incorporating both these factors gives a good fit to recognition memory data for auditory as well as visual stimuli. In addition, we present the first demonstration of the orthogonality of summed similarity and inter-item homogeneity effects. These data imply that auditory and visual representations undergo very similar transformations while they are encoded and retrieved from memory.     

Prediction of protein self-interactions using stacked long short-term memory from protein sequences information

Background

Self-interacting Proteins (SIPs) plays a critical role in a series of life function in most living cells. Researches on SIPs are important part of molecular biology. Although numerous SIPs data be provided, traditional experimental methods are labor-intensive, time-consuming and costly and can only yield limited results in real-world needs. Hence,it’s urgent to develop an efficient computational SIPs prediction method to fill the gap. Deep learning technologies have proven to produce subversive performance improvements in many areas, but the effectiveness of deep learning methods for SIPs prediction has not been verified.

Results

We developed a deep learning model for predicting SIPs by constructing a Stacked Long Short-Term Memory (SLSTM) neural network that contains “dropout”. We extracted features from protein sequences using a novel feature extraction scheme that combined Zernike Moments (ZMs) with Position Specific Weight Matrix (PSWM). The capability of the proposed approach was assessed on S.erevisiae and Human SIPs datasets. The result indicates that the approach based on deep learning can effectively resist data skew and achieve good accuracies of 95.69 and 97.88%, respectively. To demonstrate the progressiveness of deep learning, we compared the results of the SLSTM-based method and the celebrated Support Vector Machine (SVM) method and several other well-known methods on the same datasets.

Conclusion

The results show that our method is overall superior to any of the other existing state-of-the-art techniques. As far as we know, this study first applies deep learning method to predict SIPs, and practical experimental results reveal its potential in SIPs identification.

     

Recurrent neural networks of integrate-and-fire cells simulating short-term memory and wrist movement tasks derived from continuous dynamic networks

Dynamic recurrent neural networks composed of units with continuous activation functions provide a powerful tool for simulating a wide range of behaviors, since the requisite interconnections can be readily derived by gradient descent methods. However, it is not clear whether more realistic integrate-and-fire cells with comparable connection weights would perform the same functions. We therefore investigated methods to convert dynamic recurrent neural networks of continuous units into networks with integrate-and-fire cells. The transforms were tested on two recurrent networks derived by backpropagation. The first simulates a short-term memory task with units that mimic neural activity observed in cortex of monkeys performing instructed delay tasks. The network utilizes recurrent connections to generate sustained activity that codes the remembered value of a transient cue. The second network simulates patterns of neural activity observed in monkeys performing a step-tracking task with flexion/extension wrist movements. This more complicated network provides a working model of the interactions between multiple spinal and supraspinal centers controlling motoneurons. Our conversion algorithm replaced each continuous unit with multiple integrate-and-fire cells that interact through delayed "synaptic potentials". Successful transformation depends on obtaining an appropriate fit between the activation function of the continuous units and the input-output relation of the spiking cells. This fit can be achieved by adapting the parameters of the synaptic potentials to replicate the input-output behavior of a standard sigmoidal activation function (shown for the short-term memory network). Alternatively, a customized activation function can be derived from the input-output relation of the spiking cells for a chosen set of parameters (demonstrated for the wrist flexion/extension network). In both cases the resulting networks of spiking cells exhibited activity that replicated the activity of corresponding continuous units. This confirms that the network solutions obtained through backpropagation apply to spiking networks and provides a useful method for deriving recurrent spiking networks performing a wide range of functions.     

Extended Newick: it is time for a standard representation of phylogenetic networks

Background  

Phylogenetic trees resulting from molecular phylogenetic analysis are available in Newick format from specialized databases but when it comes to phylogenetic networks, which provide an explicit representation of reticulate evolutionary events such as recombination, hybridization or lateral gene transfer, the lack of a standard format for their representation has hindered the publication of explicit phylogenetic networks in the specialized literature and their incorporation in specialized databases. Two different proposals to represent phylogenetic networks exist: as a single Newick string (where each hybrid node is splitted once for each parent) or as a set of Newick strings (one for each hybrid node plus another one for the phylogenetic network).     

Learning and long- and short-term memory in cockroaches

Intact cockroaches, headless insects and isolated segments of the animal were trained to avoid electric shocks, contingent upon an electric signal which was applied a fraction of a second prior to the shocks. The results showed that all the intact insects learned to avoid the shocks in approximately ten sessions, on a schedule of two daily sessions and ten trials per session. The retention of the learned responses was further tested; it showed that there was no decrement of avoidance reactions till the end of insect life in the experimental conditions. Furthermore, this retention of the learned behaviour was not affected by the severance of the head nor even by the isolation of a single ganglion, once the behaviour had been established.The learning ability of the headless animals, and of the isolated segments, was different. There is evidence that both types of preparations demonstrated some degree of learning, but a majority of them failed to reach the criterion of 100 per cent avoidances in a training schedule, exactly the same as for the intact cockroach. Nevertheless, if the headless insects were given a mass training of fourteen sessions per day, all of them reached the criterion. The retention of the isolated segment was found to be less than 1 day long, and that of the headless insects lasted at most 2 or 3 days after the last training session.It is interpreted that long-term memory can be established in the intact animals and, once it has been accomplished, the acquired behaviour can be retained by the lower portions of the central nervous system.     

Characteristics of the formation of short-term memory in rats at various time intervals between tests

Dependency of the speed of short-term memory formation in rats on the intertrial interval was studied. The so called direct method was applied for determining the delay time of the complex perception of food location. Conclusion is made that not less than 2 minutes intertrial intervals appear to be optimal for the formation of short-term memory. Such behavioural manifestations as omission of separate trials, lengthening of the time both for grooming and running to the feeder are considered to the manifestation of brain self-regulation activity. Attention is paid to the fact that such a form of behaviour is characteristic of one group of rats and not of the other.     

Vasopressin analogues and spatial short-term memory in rats

The effect of vasopressin analogues on short-term memory was tested in the 12-arm radial maze. After the first 6 choices rats (n=6) were removed from the apparatus and allowed to complete the remaining 6 choices 20 min later. Whereas desgly-NH₂-VP, AVP, dAVP and DAVP (3.0 μ/kg) administered 40 min before or immediately after the first 6 choices did not change the incidence of errors in the second series of choices (2.0 errors under control conditions), similarly applied dDAVP deteriorated the rat's performance almost to the chance level of 3 errors. The significance of short-term memory tests for assessing the mnestic role of peptide hormones is stressed.     

Stability of dead zone bidirectional associative memory neural networks involving time delays

A mathematical model describing the dynamical interactions of bidirectional associative memory networks involving transmission delays is considered. The influence of a dead zone or a zone of noactivation on the global stability is investigated and various easily verifiable sets of sufficient conditions are established. The asymptotic nature of solutions when the given system of equations does not possess an equilibrium pattern is discussed.     

Need probability effects in animal short-term memory

Five pigeons performed in a delayed matching-to-sample (DMTS) procedure with five delay durations (0.5, 2.5, 5, 10 and 20 s) mixed within sessions. Contrary to the predictions of need probability theory, discriminability decreased when fewer short than long delays were included in each session. To test whether the decrease in discriminability was due to a decrease in obtained reinforcement at short delays, the number of trials at each delay was held constant and reinforcer probability was increased with increasing delay. This manipulation produced a similar decrease in discriminability as when the frequency of delays was manipulated. It was concluded that the effect of delay frequency on the forgetting function is mediated by the effect of the reinforcer distribution, which influences discriminability by weakening stimulus control.     

Alternative time representation in dopamine models

Dopaminergic neuron activity has been modeled during learning and appetitive behavior, most commonly using the temporal-difference (TD) algorithm. However, a proper representation of elapsed time and of the exact task is usually required for the model to work. Most models use timing elements such as delay-line representations of time that are not biologically realistic for intervals in the range of seconds. The interval-timing literature provides several alternatives. One of them is that timing could emerge from general network dynamics, instead of coming from a dedicated circuit. Here, we present a general rate-based learning model based on long short-term memory (LSTM) networks that learns a time representation when needed. Using a naïve network learning its environment in conjunction with TD, we reproduce dopamine activity in appetitive trace conditioning with a constant CS-US interval, including probe trials with unexpected delays. The proposed model learns a representation of the environment dynamics in an adaptive biologically plausible framework, without recourse to delay lines or other special-purpose circuits. Instead, the model predicts that the task-dependent representation of time is learned by experience, is encoded in ramp-like changes in single-neuron activity distributed across small neural networks, and reflects a temporal integration mechanism resulting from the inherent dynamics of recurrent loops within the network. The model also reproduces the known finding that trace conditioning is more difficult than delay conditioning and that the learned representation of the task can be highly dependent on the types of trials experienced during training. Finally, it suggests that the phasic dopaminergic signal could facilitate learning in the cortex.     

Role of rhythms in facilitating short-term memory

Working memory tasks have been associated with the appearance of elevated single unit activity (SUA) in primate studies, and oscillatory activity in the EEG or the local field potential (LFP) in humans. The study by Lee et al. in this issue of Neuron provides novel insights regarding the relationship between SUA and LFP rhythmicity in V4 during working memory tasks.     

Cortico-striatal representation of time in animals and humans

Interval timing in the seconds-to-minutes range is crucial to learning, memory, and decision-making. Recent findings argue for the involvement of cortico-striatal circuits that are optimized by the dopaminergic modulation of oscillatory activity and lateral connectivity at the level of cortico-striatal inputs. Striatal medium spiny neurons are proposed to detect the coincident activity of specific beat patterns of cortical oscillations, thereby permitting the discrimination of supra-second durations based upon the reoccurring patterns of subsecond neural firing. This proposal for the cortico-striatal representation of time is consistent with the observed psychophysical properties of interval timing (e.g. linear time scale and scalar variance) as well as much of the available pharmacological, lesion, patient, electrophysiological, and neuroimaging data from animals and humans (e.g. dopamine-related timing deficits in Huntington's and Parkinson's disease as well as related animal models). The conclusion is that although the striatum serves as a 'core timer', it is part of a distributed timing system involving the coordination of large-scale oscillatory networks.     

Visual short-term memory compared in rhesus monkeys and humans

Change detection is a popular task to study visual short-term memory (STM) in humans [1-4]. Much of this work suggests that STM has a fixed capacity of 4 ± 1 items [1-6]. Here we report the first comparison of change-detection memory between humans and a species closely related to humans, the rhesus monkey. Monkeys and humans were tested in nearly identical procedures with overlapping display sizes. Although the monkeys' STM was well fit by a one-item fixed-capacity memory model, other monkey memory tests with four-item lists have shown performance impossible to obtain with a one-item capacity [7]. We suggest that this contradiction can be resolved using a continuous-resource approach more closely tied to the neural basis of memory [8, 9]. In this view, items have a noisy memory representation whose noise level depends on display size as a result of the distributed allocation of a continuous resource. In accord with this theory, we show that performance depends on the perceptual distance between items before and after the change, and d' depends on display size in an approximately power-law fashion. Our results open the door to combining the power of psychophysics, computation, and physiology to better understand the neural basis of STM.     

Stability and attractivity in associative memory networks

We focus on stable and attractive states in a network having two-state neuron-like elements. We calculate the connection matrix which guarantees the stability and the strongest attractivity of p memorized patterns. We present an analytical evaluation of the patterns' attractivity. These results are illustrated by some computer simulations.     

Major depressive disorder diagnosis based on effective connectivity in EEG signals: a convolutional neural network and long short-term memory approach

Deep learning techniques have recently made considerable advances in the field of artificial intelligence. These methodologies can assist psychologists in early diagnosis of mental disorders and preventing severe trauma. Major Depression Disorder (MDD) is a common and serious medical condition whose exact manifestations are not fully understood. So, early discovery of MDD patients helps to cure or limit the adverse effects. Electroencephalogram (EEG) is prominently used to study brain diseases such as MDD due to having high temporal resolution information, and being a noninvasive, inexpensive and portable method. This paper has proposed an EEG-based deep learning framework that automatically discriminates MDD patients from healthy controls. First, the relationships among EEG channels in the form of effective brain connectivity analysis are extracted by Generalized Partial Directed Coherence (GPDC) and Direct directed transfer function (dDTF) methods. A novel combination of sixteen connectivity methods (GPDC and dDTF in eight frequency bands) was used to construct an image for each individual. Finally, the constructed images of EEG signals are applied to the five different deep learning architectures. The first and second algorithms were based on one and two-dimensional convolutional neural network (1DCNN–2DCNN). The third method is based on long short-term memory (LSTM) model, while the fourth and fifth algorithms utilized a combination of CNN with LSTM model namely, 1DCNN-LSTM and 2DCNN-LSTM. The proposed deep learning architectures automatically learn patterns in the constructed image of the EEG signals. The efficiency of the proposed algorithms is evaluated on resting state EEG data obtained from 30 healthy subjects and 34 MDD patients. The experiments show that the 1DCNN-LSTM applied on constructed image of effective connectivity achieves best results with accuracy of 99.24% due to specific architecture which captures the presence of spatial and temporal relations in the brain connectivity. The proposed method as a diagnostic tool is able to help clinicians for diagnosing the MDD patients for early diagnosis and treatment.     

The neural representation of time

This review summarizes recent investigations of temporal processing. We focus on motor and perceptual tasks in which crucial events span hundreds of milliseconds. One key question concerns whether the representation of temporal information is dependent on a specialized system, distributed across a network of neural regions, or computed in a local task-dependent manner. Consistent with the specialized system framework, the cerebellum is associated with various tasks that require precise timing. Computational models of timing mechanisms within the cerebellar cortex are beginning to motivate physiological studies. Emphasis has also been placed on the basal ganglia as a specialized timing system, particularly for longer intervals. We outline an alternative hypothesis in which this structure is associated with decision processes.     

Probabilistic representation of gene regulatory networks

MOTIVATION: Recent experiments have established unambiguously that biological systems can have significant cell-to-cell variations in gene expression levels even in isogenic populations. Computational approaches to studying gene expression in cellular systems should capture such biological variations for a more realistic representation. RESULTS: In this paper, we present a new fully probabilistic approach to the modeling of gene regulatory networks that allows for fluctuations in the gene expression levels. The new algorithm uses a very simple representation for the genes, and accounts for the repression or induction of the genes and for the biological variations among isogenic populations simultaneously. Because of its simplicity, introduced algorithm is a very promising approach to model large-scale gene regulatory networks. We have tested the new algorithm on the synthetic gene network library bioengineered recently. The good agreement between the computed and the experimental results for this library of networks, and additional tests, demonstrate that the new algorithm is robust and very successful in explaining the experimental data. AVAILABILITY: The simulation software is available upon request. SUPPLEMENTARY INFORMATION: Supplementary material will be made available on the OUP server.     

Vision and the representation of the surroundings in spatial memory

One of the paradoxes of vision is that the world as it appears to us and the image on the retina at any moment are not much like each other. The visual world seems to be extensive and continuous across time. However, the manner in which we sample the visual environment is neither extensive nor continuous. How does the brain reconcile these differences? Here, we consider existing evidence from both static and dynamic viewing paradigms together with the logical requirements of any representational scheme that would be able to support active behaviour. While static scene viewing paradigms favour extensive, but perhaps abstracted, memory representations, dynamic settings suggest sparser and task-selective representation. We suggest that in dynamic settings where movement within extended environments is required to complete a task, the combination of visual input, egocentric and allocentric representations work together to allow efficient behaviour. The egocentric model serves as a coding scheme in which actions can be planned, but also offers a potential means of providing the perceptual stability that we experience.