Is Novelty Detection Important in Long-Term Odor Memory?
Ontology highlight
ABSTRACT: Memory for odors is believed to be longer-lasting than memory for visual stimuli, as is evidenced by flat forgetting curves. However, performance on memory tasks is typically weaker in olfaction than vision. Studies of odor memory that use forced-choice methods confound responses that are a result of a trace memory and responses that can be obtained through process of elimination. Moreover, odor memory is typically measured with common stimuli, which are more familiar and responses may be confounded by verbal memory, and measure memory in intentional learning conditions, which are ecologically questionable. Here we demonstrate the value of using tests of memory in which hit rate and correct rejection rate are evaluated separately (i.e., not using forced-choice methods) and uncommon stimuli are used. This study compared memory for common and uncommon odors and pictures that were learned either intentionally (Exp. 1) or incidentally (Exp. 2) and tested with either a forced-choice or a one-stimulus-at-a-time ("monadic") recognition task after delays of 15 min, 48 h or 1 week. As expected, memory declined with delay in most conditions, but depended upon the particular measure of memory and was better for pictures than odors and for common than uncommon stimuli. For common odors, hit rates decreased with delay but correct rejection rates remained constant with delay. For common pictures, we found the opposite result, constant hit rates and decreased correct rejection rates. Our results support the 'misfit theory of conscious olfactory perception', which highlights the importance of the detection of novelty in olfactory memory and suggests that olfactory memory should be studied using more ecologically valid methods.
Project description:Intranasal insulin delivery is currently being used in clinical trials to test for improvement in human memory and cognition, and in particular, for lessening memory loss attributed to neurodegenerative diseases. Studies have reported the effects of short-term intranasal insulin treatment on various behaviors, but less have examined long-term effects. The olfactory bulb contains the highest density of insulin receptors in conjunction with the highest level of insulin transport within the brain. Previous research from our laboratory has demonstrated that acute insulin intranasal delivery (IND) enhanced both short- and long-term memory as well as increased two-odor discrimination in a two-choice paradigm. Herein, we investigated the behavioral and physiological effects of chronic insulin IND. Adult, male C57BL6/J mice were intranasally treated with 5?g/?l of insulin twice daily for 30 and 60days. Metabolic assessment indicated no change in body weight, caloric intake, or energy expenditure following chronic insulin IND, but an increase in the frequency of meal bouts selectively in the dark cycle. Unlike acute insulin IND, which has been shown to cause enhanced performance in odor habituation/dishabituation and two-odor discrimination tasks in mice, chronic insulin IND did not enhance olfactometry-based odorant discrimination or olfactory reversal learning. In an object memory recognition task, insulin IND-treated mice did not perform differently than controls, regardless of task duration. Biochemical analyses of the olfactory bulb revealed a modest 1.3 fold increase in IR kinase phosphorylation but no significant increase in Kv1.3 phosphorylation. Substrate phosphorylation of IR kinase downstream effectors (MAPK/ERK and Akt signaling) proved to be highly variable. These data indicate that chronic administration of insulin IND in mice fails to enhance olfactory ability, object memory recognition, or a majority of systems physiology metabolic factors - as reported to elicit a modulatory effect with acute administration. This leads to two alternative interpretations regarding long-term insulin IND in mice: 1) It causes an initial stage of insulin resistance to dampen the behaviors that would normally be modulated under acute insulin IND, but ability to clear a glucose challenge is still retained, or 2) There is a lack of behavioral modulation at high concentration of insulin attributed to the twice daily intervals of hyperinsulinemia caused by insulin IND administration without any insulin resistance, per se.
Project description:BackgroundProtein remote homology detection plays a vital role in studies of protein structures and functions. Almost all of the traditional machine leaning methods require fixed length features to represent the protein sequences. However, it is never an easy task to extract the discriminative features with limited knowledge of proteins. On the other hand, deep learning technique has demonstrated its advantage in automatically learning representations. It is worthwhile to explore the applications of deep learning techniques to the protein remote homology detection.ResultsIn this study, we employ the Bidirectional Long Short-Term Memory (BLSTM) to learn effective features from pseudo proteins, also propose a predictor called ProDec-BLSTM: it includes input layer, bidirectional LSTM, time distributed dense layer and output layer. This neural network can automatically extract the discriminative features by using bidirectional LSTM and the time distributed dense layer.ConclusionExperimental results on a widely-used benchmark dataset show that ProDec-BLSTM outperforms other related methods in terms of both the mean ROC and mean ROC50 scores. This promising result shows that ProDec-BLSTM is a useful tool for protein remote homology detection. Furthermore, the hidden patterns learnt by ProDec-BLSTM can be interpreted and visualized, and therefore, additional useful information can be obtained.
Project description:BackgroundThe accurate temporal analysis of muscle activation is of great interest in many research areas, spanning from neurorobotic systems to the assessment of altered locomotion patterns in orthopedic and neurological patients and the monitoring of their motor rehabilitation. The performance of the existing muscle activity detectors is strongly affected by both the SNR of the surface electromyography (sEMG) signals and the set of features used to detect the activation intervals. This work aims at introducing and validating a powerful approach to detect muscle activation intervals from sEMG signals, based on long short-term memory (LSTM) recurrent neural networks.MethodsFirst, the applicability of the proposed LSTM-based muscle activity detector (LSTM-MAD) is studied through simulated sEMG signals, comparing the LSTM-MAD performance against other two widely used approaches, i.e., the standard approach based on Teager-Kaiser Energy Operator (TKEO) and the traditional approach, used in clinical gait analysis, based on a double-threshold statistical detector (Stat). Second, the effect of the Signal-to-Noise Ratio (SNR) on the performance of the LSTM-MAD is assessed considering simulated signals with nine different SNR values. Finally, the newly introduced approach is validated on real sEMG signals, acquired during both physiological and pathological gait. Electromyography recordings from a total of 20 subjects (8 healthy individuals, 6 orthopedic patients, and 6 neurological patients) were included in the analysis.ResultsThe proposed algorithm overcomes the main limitations of the other tested approaches and it works directly on sEMG signals, without the need for background-noise and SNR estimation (as in Stat). Results demonstrate that LSTM-MAD outperforms the other approaches, revealing higher values of F1-score (F1-score > 0.91) and Jaccard similarity index (Jaccard > 0.85), and lower values of onset/offset bias (average absolute bias < 6 ms), both on simulated and real sEMG signals. Moreover, the advantages of using the LSTM-MAD algorithm are particularly evident for signals featuring a low to medium SNR.ConclusionsThe presented approach LSTM-MAD revealed excellent performances against TKEO and Stat. The validation carried out both on simulated and real signals, considering normal as well as pathological motor function during locomotion, demonstrated that it can be considered a powerful tool in the accurate and effective recognition/distinction of muscle activity from background noise in sEMG signals.
Project description:The present study examined why perirhinal cortex lesions in rats impair the spontaneous ability to select novel objects in preference to familiar objects, when both classes of object are presented simultaneously. The study began by repeating this standard finding, using a test of delayed object recognition memory. As expected, the perirhinal cortex lesions reduced the difference in exploration times for novel vs. familiar stimuli. In contrast, the same rats with perirhinal cortex lesions appeared to perform normally when the preferential exploration of novel vs. familiar objects was tested sequentially, i.e. when each trial consisted of only novel or only familiar objects. In addition, there was no indication that the perirhinal cortex lesions reduced total levels of object exploration for novel objects, as would be predicted if the lesions caused novel stimuli to appear familiar. Together, the results show that, in the absence of perirhinal cortex tissue, rats still receive signals of object novelty, although they may fail to link that information to the appropriate object. Consequently, these rats are impaired in discriminating the source of object novelty signals, leading to deficits on simultaneous choice tests of recognition.
Project description:Fault detection and diagnosis is one of the most critical components of preventing accidents and ensuring the system safety of industrial processes. In this paper, we propose an integrated learning approach for jointly achieving fault detection and fault diagnosis of rare events in multivariate time series data. The proposed approach combines an autoencoder to detect a rare fault event and a long short-term memory (LSTM) network to classify different types of faults. The autoencoder is trained with offline normal data, which is then used as the anomaly detection. The predicted faulty data, captured by autoencoder, are put into the LSTM network to identify the types of faults. It basically combines the strong low-dimensional nonlinear representations of the autoencoder for the rare event detection and the strong time series learning ability of LSTM for the fault diagnosis. The proposed approach is compared with a deep convolutional neural network approach for fault detection and identification on the Tennessee Eastman process. Experimental results show that the combined approach accurately detects deviations from normal behaviour and identifies the types of faults within the useful time.
Project description:This study describes a functional magnetic resonance imaging study of humans engaged in long-term memory (LTM) and working memory tasks. A pattern classifier learned to identify patterns of brain activity associated with viewing and making judgments about three categories of pictures (famous people, famous locations, and common objects). The evaluation of these stimuli relied on perception and long-term semantic and/or episodic memories. We investigated whether this classifier could successfully decode brain activity from a subsequent delayed paired-associate recognition working memory task that required the short-term retention of the same stimuli. We reasoned that the LTM-trained classifier would be able to decode delay-period activity only if that activity reflected, to some extent, the temporary activation of LTM. Our results demonstrated successful decoding: delay-period activity from a distributed network of brain regions matched learned patterns of activity for task-relevant stimuli to a greater extent than for task-irrelevant stimuli. In varying degrees throughout the delay, activity reflected the target (a retrospective code) and its associate (a prospective code) with considerable variability among subjects. Although prefrontal cortex (PFC) demonstrated category-specific patterns of activity during the LTM task, these patterns were not reinstated in PFC during the working memory task. We conclude that the short-term retention of information can be supported by the temporary reactivation of LTM representations.
Project description:BackgroundAn allergic reaction is the immune system's overreacting to a previously encountered, typically benign molecule, frequently a protein. Allergy reactions can result in rashes, itching, mucous membrane swelling, asthma, coughing, and other bizarre symptoms. To anticipate allergies, a wide range of principles and methods have been applied in bioinformatics. The sequence similarity approach's positive predictive value is very low and ineffective for methods based on FAO/WHO criteria, making it difficult to predict possible allergens.MethodThis work advocated the use of a deep learning model LSTM (Long Short-Term Memory) to overcome the limitations of traditional approaches and machine learning lower performance models in predicting the allergenicity of dietary proteins. A total of 2,427 allergens and 2,427 non-allergens, from a variety of sources, including the Central Science Laboratory and the NCBI are used. The data was divided 80:20 for training and testing purposes. These techniques have all been implemented in Python. To describe the protein sequences of allergens and non-allergens, five E-descriptors were used. E1 (hydrophilic character of peptides), E2 (length), E3(propensity to form helices), E4(abundance and dispersion), and E5 (propensity of beta strands) are used to make the variable-length protein sequence to uniform length using ACC transformation. A total of eight machine learning techniques have been taken into consideration.ResultsThe Gaussian Naive Bayes as accuracy of 64.14 %, Radius Neighbour's Classifier with 49.2 %, Bagging Classifier was 85.8 %, ADA Boost was 76.9 %, Linear Discriminant Analysis has 76.13 %, Quadratic Discriminant Analysis was 84.2 %, Extra Tree Classifier was 90%, and LSTM is 91.5 %.ConclusionAs the LSTM, has an AUC value of 91.5 % is regarded best in predicting allergens. A web server called ProAll-D has been created that successfully identifies novel allergens using the LSTM approach. Users can use the link https://doi.org/10.17632/tjmt97xpjf.1 to access the ProAll-D server and data.
Project description:The main features of grooming behavior are amazingly similar among arthropods and land vertebrates and serve the same needs. A particular pattern of cleaning movements in cockroaches shows cephalo-caudal progression. Grooming sequences become longer after adaptation to the new setting. Novelty related changes in grooming are recognized as a form of displacement behavior. Statistical analysis of behavior revealed that antennal grooming in American cockroach, Periplaneta americana L., was significantly enhanced in the presence of odor.
Project description:The interactions of L-type calcium channels (LTCCs) and NMDA receptors (NMDARs) in memories are poorly understood. Here we investigated the specific roles of anterior piriform cortex (aPC) LTCCs and NMDARs in early odor preference memory in mice. Using calcium imaging in aPC slices, LTCC activation was shown to be dependent on NMDAR activation. Either D-APV (NMDAR antagonist) or nifedipine (LTCC antagonist) reduced somatic calcium transients in pyramidal cells evoked by lateral olfactory tract stimulation. However, nifedipine did not further reduce calcium in the presence of D-APV. In mice that underwent early odor preference training, blocking NMDARs in the aPC prevented short-term (3 hr) and long-term (24 hr) odor preference memory, and both memories were rescued when BayK-8644 (LTCC agonist) was co-infused. However, activating LTCCs in the absence of NMDARs resulted in loss of discrimination between the conditioned odor and a similar odor mixture at 3 hr. Elevated synaptic AMPAR expression at 3 hr was prevented by D-APV infusion but restored when LTCCs were directly activated, mirroring the behavioral outcomes. Blocking LTCCs prevented 24 hr memory and spared 3 hr memory. These results suggest that NMDARs mediate stimulus-specific encoding of odor memory while LTCCs mediate intracellular signaling leading to long-term memory.
Project description:Increasing access to media in the 21st century has led to a rapid rise in the prevalence of media multitasking (simultaneous use of multiple media streams). Such behavior is associated with various cognitive differences, such as difficulty filtering distracting information and increased trait impulsivity. Given the rise in media multitasking by children, adolescents, and adults, a full understanding of the cognitive profile of media multitaskers is imperative. Here we investigated the relationship between chronic media multitasking and working memory (WM) and long-term memory (LTM) performance. Four key findings are reported (1) heavy media multitaskers (HMMs) exhibited lower WM performance, regardless of whether external distraction was present or absent; (2) lower performance on multiple WM tasks predicted lower LTM performance; (3) media multitasking-related differences in memory reflected differences in discriminability rather than decision bias; and (4) attentional impulsivity correlated with media multitasking behavior and reduced WM performance. These findings suggest that chronic media multitasking is associated with a wider attentional scope/higher attentional impulsivity, which may allow goal-irrelevant information to compete with goal-relevant information. As a consequence, heavy media multitaskers are able to hold fewer or less precise goal-relevant representations in WM. HMMs' wider attentional scope, combined with their diminished WM performance, propagates forward to yield lower LTM performance. As such, chronic media multitasking is associated with a reduced ability to draw on the past--be it very recent or more remote--to inform present behavior.