28 research outputs found
Author response image 1. Author response
No full textAnimals can learn causal relationships between pairs of stimuli separated in time and this ability depends on the hippocampus. Such learning is believed to emerge from alterations in network connectivity, but large-scale connectivity is difficult to measure directly, especially during learning. Here, we show that area CA1 cells converge to time-locked firing sequences that bridge the two stimuli paired during training, and this phenomenon is coupled to a reorganization of network correlations. Using two-photon calcium imaging of mouse hippocampal neurons we find that co-time-tuned neurons exhibit enhanced spontaneous activity correlations that increase just prior to learning. While time-tuned cells are not spatially organized, spontaneously correlated cells do fall into distinct spatial clusters that change as a result of learning. We propose that the spatial re-organization of correlation clusters reflects global network connectivity changes that are responsible for the emergence of the sequentially-timed activity of cell-groups underlying the learned behavior. DOI: http://dx.doi.org/10.7554/eLife.01982.001
Heterosynaptic Plasticity Underlies Aversive Olfactory Learning in Drosophila
No full textSummary Although associative learning has been localized to specific brain areas in many animals, identifying the underlying synaptic processes in vivo has been difficult. Here, we provide the first demonstration of long-term synaptic plasticity at the output site of the Drosophila mushroom body. Pairing an odor with activation of specific dopamine neurons induces both learning and odor-specific synaptic depression. The plasticity induction strictly depends on the temporal order of the two stimuli, replicating the logical requirement for associative learning. Furthermore, we reveal that dopamine action is confined to and distinct across different anatomical compartments of the mushroom body lobes. Finally, we find that overlap between sparse representations of different odors defines both stimulus specificity of the plasticity and generalizability of associative memories across odors. Thus, the plasticity we find here not only manifests important features of associative learning but also provides general insights into how a sparse sensory code is read out
Flexible specificity of memory in Drosophila depends on a comparison between choices
No full textInternational audienceMemory guides behavior across widely varying environments and must therefore be both sufficiently specific and general. A memory too specific will be useless in even a slightly different environment, while an overly general memory may lead to suboptimal choices. Animals successfully learn to both distinguish between very similar stimuli and generalize across cues. Rather than forming memories that strike a balance between specificity and generality, Drosophila can flexibly categorize a given stimulus into different groups depending on the options available. We asked how this flexibility manifests itself in the well-characterized learning and memory pathways of the fruit fly. We show that flexible categorization in neuronal activity as well as behavior depends on the order and identity of the perceived stimuli. Our results identify the neural correlates of flexible stimulus-categorization in the fruit fly
Effects of stochastic coding on olfactory discrimination in flies and mice.
No full textSparse coding can improve discrimination of sensory stimuli by reducing overlap between their representations. Two factors, however, can offset sparse coding's benefits: similar sensory stimuli have significant overlap and responses vary across trials. To elucidate the effects of these 2 factors, we analyzed odor responses in the fly and mouse olfactory regions implicated in learning and discrimination-the mushroom body (MB) and the piriform cortex (PCx). We found that neuronal responses fall along a continuum from extremely reliable across trials to extremely variable or stochastic. Computationally, we show that the observed variability arises from noise within central circuits rather than sensory noise. We propose this coding scheme to be advantageous for coarse- and fine-odor discrimination. More reliable cells enable quick discrimination between dissimilar odors. For similar odors, however, these cells overlap and do not provide distinguishing information. By contrast, more unreliable cells are decorrelated for similar odors, providing distinguishing information, though these benefits only accrue with extended training with more trials. Overall, we have uncovered a conserved, stochastic coding scheme in vertebrates and invertebrates, and we identify a candidate mechanism, based on variability in a winner-take-all (WTA) inhibitory circuit, that improves discrimination with training
Unreliable cells contribute more to discrimination of similar odors.
No full textExtended learning increases the amount of discriminatory power between odor-pairs. Odor-pairs are split into dissimilar (A, C) odor-pairs, whose correlation coefficient is ≤ 0.5. (A) Shown are 2 plots of how extended (E, x-axis) and normal (N, x-axis) training affect the contributions of (y-axis) unreliable cells (left plot, UR) and reliable cells (right plot, R) towards discrimination (termed useful Discrimination) of dissimilar odors. Each gray point represents the discrimination measure (Eq 2) for an odor-pair. Discrimination calculations for the fly are based on data from [9]. Unreliable cell contributions increase with extended training, to match the contributions of reliable cells. (B) Similar plots as (A) for similar odors. Here, too, extended training increases the discrimination contribution of unreliable (UR) cells. For reliable cells, however, contributions are maintained or reduced; (C, D) are equivalent calculations for the mouse dataset shown in Fig 1. (E, F) The improvement in discrimination from normal to extended training for similar odor-pairs in flies (E) and mice (F). Each circle represents the contribution of cells in a reliability class (x-axis) towards the ability to discriminate similar odors with extended training compared to normal training (y-axis). The highest contribution towards discrimination comes from cells with a reliability of 1, and the contributions decrease as the reliability increases. The dotted lines (y = 0) indicate the value at which there is no change in useful discrimination. Datasets used in these plots for flies (A, B, E) and mice (C, D, F) are stored within the Zenodo and Dandi repositories as the fly main dataset and mouse dataset 164 (links in Data Availability section). Additionally, plots A–C used the following supplementary fly datasets: 09042009, 110106_4, 110107_2, 110109_1, and 110109_2. The data underlying the graphs shown in the figure can be found in S5 Data.</p
Family planning success stories in Bangladesh and India
No full textThe Matlab Project in Bangladesh and the Kundam Project in India have demonstrated that a significant rise in contraceptive prevalence can occur in socioeconomic environments that are generally conducive to high fertility and mortality. The author describes the inputs and outputs of these two projects and tries to identify the factors underlying their success. Both projects are experimental in the sense that in each anintervention area is provided with special inputs that are not provided to a contiguous control area. The special inputs were different for the two projects. In the intervention area in Matlab, the project took responsibility for providing family planning and some rudimentary maternal and child health services that were considerably different from those provided in the national program. In Kundam, the project did not take responsibility for providing services in the intervention area, but rather tried to mobilize the community through various clubs and committees to take the most advantage of the government's family planning and other development programs. The success of the Matlab Project can be attributed to various aspects of the organizational system developed for delivering consumer-friendly services. The success of the Kundam Project can be attributed to various aspects of the system developed for community members'active participation in the program. The projects are not fully replicable because of inadequate human and financial resources, but the lessons learned from them should be useful in improving national programs. The Kundam Project is more realistic in the sense that it focuses on activities that supplement local activities of the national program rather than substitute for them (as in the Matlab Project). Thus the Kundam Project is more likely to be replicable than the Matlab Project.Health Monitoring&Evaluation,Adolescent Health,Reproductive Health,Early Child and Children's Health,ICT Policy and Strategies
Detecting signals with calcium imaging.
No full text(A–E) Mouse calcium imaging data registration and segmentation. (A) Top: Maximum projection GRIN lens FOV of the T-series for each plane imaged. Bar: 100 μm. Bottom: Corresponding ROIs segmented. (B) Pixel shift in the x-axis (left) and y-axis (right) during registration at each represented by frames on the x-axis. (C) Size (in pixels) distribution of each of the segmented ROIs. (D) Peak of phase correlation between frame and reference image at each time point. (E) Magnitude of the shifts (both rigid and non-rigid) for each PC of the registered movie. (F–I) Fly calcium imaging analysis. (F) A few example traces of different significant cells. The dotted line is y = 0 and indicates the value at which Ca2+ activity is 0. (G) A cumulative distribution of the base (black) and signal (red) means. The signal follows a Gamma distribution as we showed in the paper (there is a little bit of a left skew), while the base mean is a very good fit for a Gaussian. The vertical dotted line (x = 0) denotes no response. (H, I) Plot of the base (left) and signal (right) means. There are 918 cells in all. The amount of overlap is about 90 cells for both responses, and 90 of the base means are above the lowest signal mean response. And, 90 of the signal means are below the highest base mean. We took the overlapping population of cells whose base means were above the lowest signal mean response, and re-plotted their and base and signals means while joining base and signal means for the same cells. We found that in these cells, all base means are lower than their signal means. Dotted line in (H) denotes the line (x = 0) denoting no response. See the Data Availability section for details on the 2 datasets used here: the fly main dataset (F–I) on Zenodo, and mouse dataset 164 (A–E) on Dandi. The data underlying the graphs shown in the figure can be found in S6 Data. FOV, field of view; PC, principal component; ROI, region of interest.</p
Three possible models of fine-discrimination.
No full text(A) The stimulus representations in the untrained animal. For each odor, the dotted ovals enclose the cells that respond to the odor (unreliable cells are gray, reliable cells are pink). The unreliable cells are mostly non-overlapping for the 2 odors, whereas the reliable cells overlap exactly. The thickness of the arrows denotes the total strength from the 3 sets of cells to the 2 downstream neurons that drive approach or avoidance behavior. Initially, all arrows are of the same strength. (B) Saturating the cloud. With initial training (left), the strength of downstream synapses of reliable cells saturates (i.e., weakens to near zero), but synapses of unreliable cells barely change their strength. Consequently, as reliable cell overlap is high between the 2 odors, approach/avoid neurons receive equal input, and the animal is unable to discriminate. With extended training (right), the downstream synapses of unreliable cells also saturate, driving the 2 stimuli apart, as depicted by the thicker arrows. The connection strengths to the approach MBON are much weaker than to the avoid MBON, and therefore odor A and B are distinguishable. (C) Resize the cloud. The stimulus representation shrinks to only include cells that are non-overlapping between the 2 stimuli (e.g., the shared reliable cells are turned off). While some resizing may occur with normal training, with extended training, the 2 stimuli are further driven apart, enabling the animal to discriminate the 2 odors. (D) Move the cloud. Some mechanism (most likely driven by unreliable cells, the cells that are most unlike between similar odors) rearranges the stimulus ensemble response to include new cells, or converts unreliable to reliable cells that do not overlap, and thus improve discrimination. MBON, MB output neuron.</p
Cell reliability, response size, and odor specificity follow a continuous distribution.
No full text(A) Schematic of 3 response properties that were examined for odors and odor-pairs. Odor reliability refers to the number of trials for which a cell is responsive. Each trial is depicted by a grid square, and black color indicates a response. Cells that respond in more than half the trials were classified as reliable; all other responsive cells were unreliable. Odor response size is the firing rate of the cell in individual trials, as measured by Ca2+ fluorescence. The top line shows a cell with a large response, where the intensity of gray denotes size of response. Odor overlap is the probability that the cell will respond to both odors. The odor overlap here is shown in yellow and decreases for dissimilar odors (bottom) compared to similar odors (top). (B, E) Cumulative frequency histograms showing the distribution of reliabilities for KCs (flies) and PCx cells (mice). Each circle represents the cumulative probability for cells with a reliability value represented on the x-axis. The points are fit by Gamma distributions with shape = 0.64, scale = 0.42 (for B, fly) and shape = 0.64, scale = 0.17 (for E, mice). The significance of the Gamma distribution is that it is a maximum entropy code and optimizes for the most stimuli that can be encoded (see text). (C, F) Cumulative frequency histograms for response sizes in both flies and mice are well fit by Gamma distributions with shape = 0.77, scale = 0.28 (for C, fly) and shape = 0.70, scale = 0.24 (for F, mice). (D, G) Cumulative frequency histograms for the overlap of cells between odor-pairs in both flies and mice, also fit using Gamma distributions with shape = 0.19, scale = 0.60 (for D, fly) and shape = 0.18, scale = 0.35 (for G, mice). Datasets used in these plots for flies (B–D) and mice (E–G) are stored within the Zenodo and Dandi repositories as the main dataset for flies and dataset 164 for mice, respectively. The data underlying the graphs shown in the figure can be found in S2 Data. KC, Kenyon cell; PCx, piriform cortex.</p
Data that underlies Fig 4.
No full textSparse coding can improve discrimination of sensory stimuli by reducing overlap between their representations. Two factors, however, can offset sparse coding’s benefits: similar sensory stimuli have significant overlap and responses vary across trials. To elucidate the effects of these 2 factors, we analyzed odor responses in the fly and mouse olfactory regions implicated in learning and discrimination—the mushroom body (MB) and the piriform cortex (PCx). We found that neuronal responses fall along a continuum from extremely reliable across trials to extremely variable or stochastic. Computationally, we show that the observed variability arises from noise within central circuits rather than sensory noise. We propose this coding scheme to be advantageous for coarse- and fine-odor discrimination. More reliable cells enable quick discrimination between dissimilar odors. For similar odors, however, these cells overlap and do not provide distinguishing information. By contrast, more unreliable cells are decorrelated for similar odors, providing distinguishing information, though these benefits only accrue with extended training with more trials. Overall, we have uncovered a conserved, stochastic coding scheme in vertebrates and invertebrates, and we identify a candidate mechanism, based on variability in a winner-take-all (WTA) inhibitory circuit, that improves discrimination with training.</div
