Backpropagation and the brain.

During learning, the brain modifies synapses to improve behaviour. In the cortex, synapses are embedded within multilayered networks, making it difficult to determine the effect of an individual synaptic modification on the behaviour of the system. The backpropagation algorithm solves this problem in deep artificial neural networks, but historically it has been viewed as biologically problematic. Nonetheless, recent developments in neuroscience and the successes of artificial neural networks have reinvigorated interest in whether backpropagation offers insights for understanding learning in the cortex. The backpropagation algorithm learns quickly by computing synaptic updates using feedback connections to deliver error signals. Although feedback connections are ubiquitous in the cortex, it is difficult to see how they could deliver the error signals required by strict formulations of backpropagation. Here we build on past and recent developments to argue that feedback connections may instead induce neural activities whose differences can be used to locally approximate these signals and hence drive effective learning in deep networks in the brain.

Adult neurogenesis acts as a neural regularizer.

New neurons are continuously generated in the subgranular zone of the dentate gyrus throughout adulthood. These new neurons gradually integrate into hippocampal circuits, forming new naive synapses. Viewed from this perspective, these new neurons may represent a significant source of "wiring" noise in hippocampal networks. In machine learning, such noise injection is commonly used as a regularization technique. Regularization techniques help prevent overfitting training data and allow models to generalize learning to new, unseen data. Using a computational modeling approach, here we ask whether a neurogenesis-like process similarly acts as a regularizer, facilitating generalization in a category learning task. In a convolutional neural network (CNN) trained on the CIFAR-10 object recognition dataset, we modeled neurogenesis as a replacement/turnover mechanism, where weights for a randomly chosen small subset of hidden layer neurons were reinitialized to new values as the model learned to categorize 10 different classes of objects. We found that neurogenesis enhanced generalization on unseen test data compared to networks with no neurogenesis. Moreover, neurogenic networks either outperformed or performed similarly to networks with conventional noise injection (i.e., dropout, weight decay, and neural noise). These results suggest that neurogenesis can enhance generalization in hippocampal learning through noise injection, expanding on the roles that neurogenesis may have in cognition.

Efficacy of Waterproof Cast Protectors and Their Ability to Keep Casts Dry.

PURPOSE: The purpose of the study was to compare the efficacy of 6 different commercially available waterproof cast protectors in their ability to maintain a dry environment and evaluate whether cast protectors perform better than a plastic bag secured with tape in keeping casts dry. METHODS: We enrolled 23 adult participants to test 6 different commercially available cast protectors and a plastic bag. Participants trialed all cast protectors twice, with and without motion, by fully submerging each cast protector in water with a paper towel held between their index and middle fingers. Moisture accumulation within the cast protectors was estimated by the change in weight of paper towel. An analysis of variance test was performed to compare differences between cast protectors in their ability to maintain a dry internal environment. RESULTS: The plastic bag showed an average moisture accumulation of 5.50 g without motion compared with all other cast protectors, which had 0.0 g of moisture accumulation. One cast protector and the plastic bag had an average moisture accumulation of 0.46 g and 4.51 g with motion compared to all other cast protectors. The plastic bag was ranked the worst by 100% of participants. CONCLUSIONS: Cast protectors appear to offer superior protection from moisture compared with a plastic bag. TYPE OF STUDY/LEVEL OF EVIDENCE: Therapeutic IV.

Complications after operative treatment of high-grade acromioclavicular injuries.

BACKGROUND: Treatment of Rockwood type III-V acromioclavicular (AC) joint injuries remains controversial. Numerous reconstruction techniques have been proposed. The purpose of this study was to describe the complication profile in a large cohort of patients who underwent surgical management of AC joint separations using a variety of reconstruction strategies. METHODS: All patients who underwent surgery for AC joint injuries from 2013 to 2019 at a single institution were identified. Chart review was performed to capture patient demographics, radiographic measurements, operative techniques, postoperative complications, and revision surgery. Structural failure was defined as a radiographic loss of reduction of greater than 50% when comparing immediate and final postoperative imaging. Logistic regression analysis was performed to identify risk factors for complications and revision surgery. RESULTS: Of the 279 patients included in the study, 66 (24%) had type III separations, 20 (7%) type IV, and 193 (69%) type V. Fifty-three percent underwent surgery acutely (<6 weeks of injury), and 40% were chronic (>3 months). A total of 252 of the 279 surgeries (90%) were performed open and 27 (10%) were arthroscopically assisted. Allograft was used in 164 of 279 (59%) cases. Specific operative techniques (with or without allograft) included hook plating (1%), modified Weaver Dunn (16%), cortical button fixation (18%), and suture fixation (65%). At mean 28-week follow-up, 108 complications were identified in 97 patients (35%). Complications were identified at mean 20 ± 21 weeks. Sixty-nine structural failures (25%) were identified. Persistent AC joint pain requiring injection, clavicle fracture, adhesive capsulitis, and hardware complications were the other most common complications. Twenty-one patients (8%) underwent unplanned revision surgery at mean 38 ± 28 weeks after the index procedure, with the most common indication for structural failure, hardware complications, or fracture of the clavicle or coracoid. Patients who had surgery after 6 weeks from injury had a significantly greater risk of having a complication (odds ratio [OR] 3.19, 95% confidence interval [CI] 1.34-7.77, P = .009) and a significantly greater risk of having structural failure (OR 2.65, 95% CI 1.38-5.28, P = .004). Patients who had an arthroscopic technique had a greater risk of structural failure (P = .002). Complications, structural failure, and revision surgery were not significantly correlated with use of allograft or specific operative techniques. CONCLUSION: Surgical management of AC joint injuries is associated with a relatively high complication profile. Loss of reduction in the postoperative period is common. However, the revision surgery rate is low. These findings are important for patient preoperative counseling.

Publishing fast and slow: A path toward generalizability in psychology and AI.

Artificial intelligence (AI) shares many generalizability challenges with psychology. But the fields publish differently. AI publishes fast, through rapid preprint sharing and conference publications. Psychology publishes more slowly, but creates integrative reviews and meta-analyses. We discuss the complementary advantages of each strategy, and suggest that incorporating both types of strategies could lead to more generalizable research in both fields.

Elevation of Hippocampal Neurogenesis Induces a Temporally Graded Pattern of Forgetting of Contextual Fear Memories.

Throughout life neurons are continuously generated in the subgranular zone of the hippocampus. The subsequent integration of newly generated neurons alters patterns of dentate gyrus input and output connectivity, potentially rendering memories already stored in those circuits harder to access. Consistent with this prediction, we previously showed that increasing hippocampal neurogenesis after training induces forgetting of hippocampus-dependent memories, including contextual fear memory. However, the brain regions supporting contextual fear memories change with time, and this time-dependent memory reorganization might regulate the sensitivity of contextual fear memories to fluctuations in hippocampal neurogenesis. By virally expressing the inhibitory designer receptor exclusively activated by designer drugs, hM4Di, we first confirmed that chemogenetic inhibition of dorsal hippocampal neurons impairs retrieval of recent (day-old) but not remote (month-old) contextual fear memories in male mice. We then contrasted the effects of increasing hippocampal neurogenesis at recent versus remote time points after contextual fear conditioning in male and female mice. Increasing hippocampal neurogenesis immediately following training reduced conditioned freezing when mice were replaced in the context 1 month later. In contrast, when hippocampal neurogenesis was increased time points remote to training, conditioned freezing levels were unaltered when mice were subsequently tested. These temporally graded forgetting effects were observed using both environmental and genetic interventions to increase hippocampal neurogenesis. Our experiments identify memory age as a boundary condition for neurogenesis-mediated forgetting and suggest that, as contextual fear memories mature, they become less sensitive to changes in hippocampal neurogenesis levels because they no longer depend on the hippocampus for their expression.SIGNIFICANCE STATEMENT New neurons are generated in the hippocampus throughout life. As they integrate into the hippocampus, they remodel neural circuitry, potentially making information stored in those circuits harder to access. Consistent with this, increasing hippocampal neurogenesis after learning induces forgetting of the learnt information. The current study in mice asks whether these forgetting effects depend on the age of the memory. We found that post-training increases in hippocampal neurogenesis only impacted recently acquired, and not remotely acquired, hippocampal memories. These experiments identify memory age as a boundary condition for neurogenesis-mediated forgetting, and suggest remote memories are less sensitive to changes in hippocampal neurogenesis levels because they no longer depend critically on the hippocampus for their expression.

Structural foundations of optogenetics: Determinants of channelrhodopsin ion selectivity.

The structure-guided design of chloride-conducting channelrhodopsins has illuminated mechanisms underlying ion selectivity of this remarkable family of light-activated ion channels. The first generation of chloride-conducting channelrhodopsins, guided in part by development of a structure-informed electrostatic model for pore selectivity, included both the introduction of amino acids with positively charged side chains into the ion conduction pathway and the removal of residues hypothesized to support negatively charged binding sites for cations. Engineered channels indeed became chloride selective, reversing near -65 mV and enabling a new kind of optogenetic inhibition; however, these first-generation chloride-conducting channels displayed small photocurrents and were not tested for optogenetic inhibition of behavior. Here we report the validation and further development of the channelrhodopsin pore model via crystal structure-guided engineering of next-generation light-activated chloride channels (iC++) and a bistable variant (SwiChR++) with net photocurrents increased more than 15-fold under physiological conditions, reversal potential further decreased by another ∼ 15 mV, inhibition of spiking faithfully tracking chloride gradients and intrinsic cell properties, strong expression in vivo, and the initial microbial opsin channel-inhibitor-based control of freely moving behavior. We further show that inhibition by light-gated chloride channels is mediated mainly by shunting effects, which exert optogenetic control much more efficiently than the hyperpolarization induced by light-activated chloride pumps. The design and functional features of these next-generation chloride-conducting channelrhodopsins provide both chronic and acute timescale tools for reversible optogenetic inhibition, confirm fundamental predictions of the ion selectivity model, and further elucidate electrostatic and steric structure-function relationships of the light-gated pore.

Is coding a relevant metaphor for building AI?

Brette contends that the neural coding metaphor is an invalid basis for theories of what the brain does. Here, we argue that it is an insufficient guide for building an artificial intelligence that learns to accomplish short- and long-term goals in a complex, changing environment.

Memory Transformation Enhances Reinforcement Learning in Dynamic Environments.

Over the course of systems consolidation, there is a switch from a reliance on detailed episodic memories to generalized schematic memories. This switch is sometimes referred to as "memory transformation." Here we demonstrate a previously unappreciated benefit of memory transformation, namely, its ability to enhance reinforcement learning in a dynamic environment. We developed a neural network that is trained to find rewards in a foraging task where reward locations are continuously changing. The network can use memories for specific locations (episodic memories) and statistical patterns of locations (schematic memories) to guide its search. We find that switching from an episodic to a schematic strategy over time leads to enhanced performance due to the tendency for the reward location to be highly correlated with itself in the short-term, but regress to a stable distribution in the long-term. We also show that the statistics of the environment determine the optimal utilization of both types of memory. Our work recasts the theoretical question of why memory transformation occurs, shifting the focus from the avoidance of memory interference toward the enhancement of reinforcement learning across multiple timescales. SIGNIFICANCE STATEMENT: As time passes, memories transform from a highly detailed state to a more gist-like state, in a process called "memory transformation." Theories of memory transformation speak to its advantages in terms of reducing memory interference, increasing memory robustness, and building models of the environment. However, the role of memory transformation from the perspective of an agent that continuously acts and receives reward in its environment is not well explored. In this work, we demonstrate a view of memory transformation that defines it as a way of optimizing behavior across multiple timescales.

Building machines that learn and think for themselves.

We agree with Lake and colleagues on their list of "key ingredients" for building human-like intelligence, including the idea that model-based reasoning is essential. However, we favor an approach that centers on one additional ingredient: autonomy. In particular, we aim toward agents that can both build and exploit their own internal models, with minimal human hand engineering. We believe an approach centered on autonomous learning has the greatest chance of success as we scale toward real-world complexity, tackling domains for which ready-made formal models are not available. Here, we survey several important examples of the progress that has been made toward building autonomous agents with human-like abilities, and highlight some outstanding challenges.

When Can I Drive After Orthopaedic Surgery? A Systematic Review.

BACKGROUND: Patients often ask their doctors when they can safely return to driving after orthopaedic injuries and procedures, but the data regarding this topic are diverse and sometimes conflicting. Some studies provide observer-reported outcome measures, such as brake response time or simulators, to estimate when patients can safely resume driving after surgery, and patient survey data describing when patients report a return to driving, but they do not all agree. We performed a systematic review and quality appraisal for available data regarding when patients are safe to resume driving after common orthopaedic surgeries and injuries affecting the ability to drive. QUESTIONS/PURPOSES: Based on the available evidence, we sought to determine when patients can safely return to driving after (1) lower extremity orthopaedic surgery and injuries; (2) upper extremity orthopaedic surgery and injuries; and (3) spine surgery. METHODS: A search was performed using PubMed and EMBASE®, with a list of 20 common orthopaedic procedures and the words "driving" and "brake". Selection criteria included any article that evaluated driver safety or time to driving after major orthopaedic surgery or immobilization using observer-reported outcome measures or survey data. A total of 446 articles were identified from the initial search, 48 of which met inclusion criteria; abstract-only publications and non-English-language articles were not included. The evidence base includes data for driving safety on foot, ankle, spine, and leg injuries, knee and shoulder arthroscopy, hip and knee arthroplasty, carpal tunnel surgery, and extremity immobilization. Thirty-four of the articles used observer-reported outcome measures such as total brake time, brake response time, driving simulator, and standardized driving track results, whereas the remaining 14 used survey data. RESULTS: Observer-reported outcome measures of total brake time, brake response time, and brake force postoperatively suggested patients reached presurgical norms 4 weeks after right-sided procedures such as TKA, THA, and ACL reconstruction and approximately 1 week after left-sided TKA and THA. The collected survey data suggest patients resumed driving 1 month after right-sided and left-sided TKAs. Patients who had THA reported returning to driving between 6 days and 3 months postoperatively. Observer-reported outcome measures showed that patients' driving abilities often are impaired when wearing an immobilizing cast above or below the elbow or a shoulder sling on their dominant arm. Patients reported a return to driving on average 2 months after rotator cuff repair procedures and approximately 1-3 months postoperatively for total shoulder arthroplasties. Most patients with spine surgery had normal brake response times at the time of hospital discharge. Patients reported driving 6 weeks after total disc arthroplasty and anterior cervical discectomy and fusion procedures. CONCLUSIONS: The available evidence provides a best-case scenario for when patients can return to driving. It is important for observer-reported outcome measures to have normalized before a patient can consider driving, but other factors such as strength, ROM, and use of opioid analgesics need to be considered. This review can provide a guideline for when physicians can begin to consider evaluating these other factors and discussing a return to driving with patients. Survey data suggest that patients are returning to driving before observer-reported outcome measures have normalized, indicating that physicians should tell patients to wait longer before driving. Further research is needed to correlate observer-reported outcome measures with adverse events, such as motor vehicle accidents, and clinical tests that can be performed in the office. LEVEL OF EVIDENCE: Level III, therapeutic study.

Low acromial insufficiency fracture rate in reverse shoulder arthroplasty with distal clavicle excision.

Background: This study investigated the rate of acromial insufficiency fractures (AIF) in patients undergoing reverse shoulder arthroplasty (RSA) with concomitant distal clavicle excision (DCE). Methods: Patients who underwent primary RSA with DCE by a single surgeon from 2010 to 2021 were identified. Exclusion criteria included revision RSA, RSA for fracture, or cases utilizing an augmented baseplate or bone graft. AIF was defined as a radiographically proven acromion or scapular spine fracture. Pain without an identifiable fracture on imaging was defined as an acromial insufficiency reaction. Patient demographics, implant information, and radiograph measurements were compared between patients with and without acromial pathology. Results: One hundred and seventy-five patients were included. Mean age was 72.8 years, and 67% of patients were female. There were 3/174 acromial insufficiency fractures (1.7%). AIF occurred at a mean of 9.3 months after surgery. Twelve patients had insufficiency reactions (6.9%). Patients with acromial pathology were more likely to be female (p = .003) and have a diagnosis of osteoporosis (p = .047) and inflammatory arthritis (p = .049). There was no significant difference between groups in terms of other factors. Conclusion: The AIF rate in patients who underwent RSA with DCE was 1.7%. These findings suggest that DCE in the setting of RSA may have a protective role against AIF.

Large Femoral Heads With Small Acetabular Components: An Examination of 10-Year Polyethylene Wear.

Highly cross-linked polyethylene (HXLPE) has improved polyethylene (PE) wear rates while decreasing osteolysis and aseptic loosening. However, concerns exist regarding the risk of mechanical failure with thin HXLPE liners in total hip arthroplasty (THA). Our purpose was to evaluate long-term outcomes and PE wear rates in primary THAs pairing large femoral heads with small acetabular components and thin HXLPE liners. We retrospectively reviewed 29 patients who underwent 33 primary THAs using large femoral heads (32 or 36 mm) with small acetabular components (48 or 50 mm) and thin HXLPE liners (3.9 or 5.9 mm) at minimum 10-year follow-up. PE liner wear was measured using a validated radiographic technique. Mean age was 66 years, 97% of the patients were women, and mean body mass index was 26.3 kg/m2. Thirty hips (90.9%) had ceramic femoral heads, and 13 hips (39%) had 36-mm femoral heads with 3.9-mm HXLPE liners. All cases used a neutral PE design. Mean linear wear rate and volumetric wear rate were 0.04 mm/year and 39.6 mm3/year, respectively, at mean 10.9-year follow-up. There were no instances of liner fracture, liner dissociation, or revision. Mean Hip disability and Osteoarthritis Outcome Score for Joint Replacement (HOOS, JR) was 92.1. In the largest long-term study of primary THAs using large femoral heads with small acetabular components and thin HXLPE liners, we found low linear and volumetric wear rates and no cases of liner mechanical failures. Thin HXLPE liners are a safe and viable option for THA surgeons. [Orthopedics. 2024;47(2):118-122.].

A Role for Outpatient Total Joint Arthroplasty During the COVID-19 Pandemic.

Outpatient total joint arthroplasty (TJA) gained favor during the COVID-19 pandemic. Our objective was to critically evaluate the safety and short-term outcomes of outpatient total hip arthroplasty (THA) and total knee arthroplasty (TKA) throughout the pandemic compared with pre-pandemic given their increased use. This was a retrospective review of 941 patients who underwent elective, outpatient TKA and THA at a single institution. Patients were divided into two cohorts: pre-pandemic (N=611) and pandemic (N=330). Data points included demographics, comorbidities, failure of same calendar day discharge (SCDD), and 90-day readmissions and re-operations. The pandemic cohort was older (61.0 vs 59.0 years; P=.001) and had a higher Charlson Comorbidity Index (2 vs 1; P=.004). There were no differences between the pandemic and pre-pandemic cohorts regarding SCDD failures (23.3% vs 21.9%; P=.623) and reoperations (4.2% vs 6.2%; P=.205). The pandemic cohort had a lower readmission rate (4.8% vs 8.2%; P=.056). Overall, patients who had failure of SCDD more commonly underwent TKA than THA (63.5% vs 36.5%; P<.001), were older (63.0 vs 58.5 years; P<.001), had a higher body mass index (30.8 vs 29.9 kg/m2; P=.006) and Charlson Comorbidity Index (2 vs 1; P<.001), had an American Society of Anesthesiologists class greater than 2 (51.2% vs 31.6%; P<.001), received physical therapy later postoperatively (4.15 vs 3.12 hours; P<.001), and had a history significant for myocardial infarction (P=.025), chronic kidney disease (P=.004), or diabetes (P=.003). This study supports broadening indications for outpatient TJA because we found a shift toward older patients with more comorbidities successfully undergoing the procedure. [Orthopedics. 2023;46(5):303-309.].

Patient-Specific E-mailed Discharge Instructions Improve Patient Satisfaction and Patient Understanding After Surgical Arthroscopy.

Purpose: The purpose of this study is to determine whether patient-specific e-mails after surgical arthroscopy improve patient satisfaction and patient understanding of their procedure compared to traditional, preprinted discharge instructions. Methods: Sixty patients who underwent surgical arthroscopy were prospectively, randomized into two separate groups. One cohort received a detailed e-mail of their procedure, discharge instructions, and labeled intraoperative arthroscopic images, while the second cohort received the standard preprinted instructions, while their arthroscopic images were discussed at the time of follow-up. The procedures were performed by a single surgeon. All patients were seen at 1-week follow-up and given a 14-question survey specific to their postoperative course, discharge instructions, and overall satisfaction using a 5-point Likert Scale. Demographic information was collected and data points comparing overall patient satisfaction, ease of understanding instructions, quality of information, and the number of times referenced were analyzed using nonparametric tests between the two cohorts. Results: Patients in the e-mail cohort were significantly more satisfied with their surgery than patients in the printed cohort (medians: 5 versus 4, Wilcoxon chi-square = 9.98; P =.002). Patients in the e-mail cohort indicated that their instructions more greatly enhanced their overall understanding of their surgery (medians: 5 vs 3, Wilcoxon chi-square = 10.84; P = .001) and were more helpful to their recovery (medians: 5 vs 3, Wilcoxon chi-square = 7.37; P = .007). E-mail patients were significantly more likely to recommend similar instructions be sent to a friend undergoing surgery (medians: 5 versus 3, Wilcoxon chi-square = 11.10; P < .001) and share their instructions with others 72% (18/25) versus 34.5% (10/29). There was no significant difference between the e-mail cohort and the print cohort for the number of times patients referred to their instructions (medians: 3 versus 3, Wilcoxon chi-square = 2.41; P =.121). Conclusions: Patient-specific e-mailed discharge instructions improve patient satisfaction and overall understanding of the procedure compared with traditional printed discharge instructions after surgical arthroscopy. Level of Evidence: Level II, prospective randomized trial.

Midterm outcomes and survivorship of arthroscopic elbow debridement: a comparison of posttraumatic versus primary degenerative osteoarthritis.

BACKGROUND: Arthroscopic debridement is an effective means of surgical management of both degenerative osteoarthritis (DOA) and posttraumatic arthritis (PTA) of the elbow. However, the difference in the efficacy and longevity of this procedure when performed for these two distinct pathologies remains in question. The purpose of this study was to identify and compare the midterm outcomes and survivorship of arthroscopic debridement of elbow PTA and DOA. METHODS: A retrospective analysis of patients undergoing arthroscopic debridement of DOA and PTA of the elbow was performed. A questionnaire containing the Oxford Elbow Score, as well as questions regarding the incidence of reoperation, additional nonoperative intervention, complications, pain, and satisfaction, was given at 5 years, minimum, after surgery. The midterm survivorship of arthroscopic debridement free of reoperation for any reason, as well as the remaining outcome measurements obtained via the questionnaire and in-office evaluation, was compared between PTA and DOA cohorts. RESULTS: Eighty patients (DOA = 36, PTA = 44) were included in this study for analysis. All 36 patients with DOA were noted to be male. Follow-up time at the date of questionnaire response was 7.9 years (range, 5.6-11.8) in the DOA cohort and 8.6 years (range, 5.7-12.7) in the PTA cohort. Reoperation rates of 5.6% and 11.4% were identified in the DOA and PTA cohorts, respectively. No statistical difference was noted in reoperation rate, survivorship, or any measured patient-reported outcomes between cohorts at the final follow-up visit. Both cohorts demonstrated a significant improvement in Visual Analog Scale pain scores (P < .001) and ROM. Postoperative ROM was obtained at the final clinic visit at an average follow-up duration of 151 days and 255 days in the DOA and PTA cohorts, respectively. However, no difference in the degree of improvement in either outcome variable was identified after a comparison between cohorts. CONCLUSION: Arthroscopic debridement is an equally efficacious treatment option for DOA and PTA of the elbow. Patients with either pathology can expect satisfactory elbow function and an improvement in pain with little chance of reoperation at the midterm of the follow-up duration.

Backpropagation through time and the brain.

It has long been speculated that the backpropagation-of-error algorithm (backprop) may be a model of how the brain learns. Backpropagation-through-time (BPTT) is the canonical temporal-analogue to backprop used to assign credit in recurrent neural networks in machine learning, but there's even less conviction about whether BPTT has anything to do with the brain. Even in machine learning the use of BPTT in classic neural network architectures has proven insufficient for some challenging temporal credit assignment (TCA) problems that we know the brain is capable of solving. Nonetheless, recent work in machine learning has made progress in solving difficult TCA problems by employing novel memory-based and attention-based architectures and algorithms, some of which are brain inspired. Importantly, these recent machine learning methods have been developed in the context of, and with reference to BPTT, and thus serve to strengthen BPTT's position as a useful normative guide for thinking about temporal credit assignment in artificial and biological systems alike.

Chasing the trace.

Event memories are stored in hippocampal-cortical networks. In this issue of Neuron, two studies, Cowansage et al. (2014) and Tanaka et al. (2014), tag active cells during memory encoding and optogenetically manipulate the activity of these "engram" cells during subsequent recall to reveal how hippocampal and cortical cell ensembles interact during retrieval.

Patterns across multiple memories are identified over time.

Memories are not static but continue to be processed after encoding. This is thought to allow the integration of related episodes via the identification of patterns. Although this idea lies at the heart of contemporary theories of systems consolidation, it has yet to be demonstrated experimentally. Using a modified water-maze paradigm in which platforms are drawn stochastically from a spatial distribution, we found that mice were better at matching platform distributions 30 d compared to 1 d after training. Post-training time-dependent improvements in pattern matching were associated with increased sensitivity to new platforms that conflicted with the pattern. Increased sensitivity to pattern conflict was reduced by pharmacogenetic inhibition of the medial prefrontal cortex (mPFC). These results indicate that pattern identification occurs over time, which can lead to conflicts between new information and existing knowledge that must be resolved, in part, by computations carried out in the mPFC.