https://williamcoon.info/about daily 1.0 2026-03-14 https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1538095971974-FIW5NHQY5J0JAMFU5MPU/42438039_10103798231121317_7674709127448231936_n.jpg About https://williamcoon.info/contact daily 0.75 2026-03-13 https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/3aca5e1c-b9fe-4648-8e48-b26c16d70259/jhu_logo_bw.png Contact https://williamcoon.info/tools daily 0.75 2026-03-14 https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1538410277605-SYQ4LVXTQ4L21PHI7N3R/default_0_16x9-01%2Bcopy.png Tools https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/6443b80a-cd1d-44f5-b54e-6632899541c3/149156322.jpeg Tools - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1538094834539-4CN5IFGLD4THUHFYLGRU/Screen+Shot+2018-09-27+at+8.32.55+PM.png Tools https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1538094442071-CRQ477TI83Z7704VPCB1/1_qllYdaKrZLMhK-nK1zUVEQ.png Tools https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1536499822456-0XR81MC0K1O4HGCFDBXL/Screen+Shot+2018-09-02+at+10.02.56+AM.png Tools https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1554817003642-KWNBVUS9NIEG46X0CA9G/splash.png Tools https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1535903804447-OHEUAZ7FXQA8S0V1MP8F/logo_neuralact.png Tools https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1536497282305-L7FL9SYBB3FXXE91MN3T/fscortex.png Tools https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1536497415534-UKXY7Y09X5DEA38NZ6VN/spm2.png Tools https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1536498209951-NKBJD1UTVQD9M9L4H1UG/download.png Tools https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1536498844905-DZOV5QW7RUNUYYSRZ1DR/wilber-big.png Tools https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1536499057878-LLSV56CKDU2304XY0LN2/download-2.jpg Tools https://williamcoon.info/talks daily 0.75 2021-09-11 https://williamcoon.info/research daily 0.75 2026-03-13 https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/bec54737-9e2c-4a36-bf6b-804c7aec8235/SLEEP2.0-FIGS+%5BRecovered%5D-21+copy.png Research - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/7e531c7f-eae1-44c2-8742-6f4dbafb572c/Screen+Shot+2022-09-29+at+10.06.58+AM.png Research - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/c8e8265f-087e-467a-bdee-35eb02301a9a/zmax-montage-art-aware.png Research - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/9e6874bf-8f30-4ffb-9b21-0a65a8e07fb0/dcm_head.png Research - DCMini - Miniaturized Biopotential Amplifier DCMini is a miniaturized biopotential amplifier and multi-sensor system designed for research applications in brain-computer interfaces (BCI), muscular interfaces, and closed-loop biofeedback. It combines a flexible set of sensors with powerful wireless telemetry and SD logging — in a package you can stick on your head. https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/56180536-f28c-498d-bd90-5e0794d67ec1/Screen+Shot+2022-09-29+at+9.47.36+AM.png Research - Make it stand out Continual Learning in A.I. benefits from three sleep mechanisms: NREM, REM and Synaptic Downscaling (arXiv preprint here). In A, we see a conceptual visualization of sleep’s synaptic downscaling process — a size-dependent reduction in the magnitude of synaptic connections or weights, thought to be responsible in biological brains for metabolic maintenance, neural upkeep, and fine-tuning signal-to-noise ratios in neural networks. B and C show the effect of introducing downscaling to an artificial neural network performing a sequential learning task. The network undergoes interleaved periods of veridicial memory replay (akin to NREM sleep) and generative memory replay (akin to REM sleep; C, red bar), or those processes plus synaptic downscaling (C, blue bar). Looking at the distribution of learned data represented by the colors inside each bar in C, we observe that the tripartite sleep model causes a re-balancing of learning, recovering data learned in the distant past while allowing new data to be learned in the future. The waterfall plots on the right show the same effect, but as a function of “time”, showing how this rebalancing evolves over the course of the network’s “sleep”. https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1552845751758-POXH373826FWDT3LELUO/CrossNetworkSingleTrialTrajectoryFig_v4-01__.png Research Fig. 2 (from Coon et al. 2016). Exemplary spatiotemporal trajectory of task-related neuronal population activity in one single trial. A) Task-related cortical locations from Subject A (left), and the time course of neuronal population activity in exemplary locations (right). Increases in activity at each location are brief. B) Population-level activity occurs during the trough of alpha oscillations (8–12 Hz) https://images.squarespace-cdn.com/content/v1/5b8323951137a648e0f4e049/1552846522288-PA11ANVVS7RM7M1Z4NW8/Detector_Transparent-01.png Research Fig. 4 (from Coon & Schalk 2016). Single-trial detection framework for determining where and when task-related neuronal population activity occurs. (A) Analysis of signal variance (see Methods) in each channel, time-locked to stimulus onset and averaged across trials, identifies locations with activity that was modulated by the task. Task-related locations will exhibit a total variance in the signal much greater than the variance of binned subsections of the same time series (ratio ≫1; bars depict variances). In locations not related to the task, these variances are approximately the same (ratio ≈ 1). Task-related locations are defined as those with statistically significant activation ratios ((B); see text for statistics). (C) The time course of broad- band activity in one trial from one location. For each location, the detection algorithm optimizes a channel-specific threshold by scanning through a series of values applied to all trials ((D) and selecting the threshold value that maximizes the difference between the number of hits detected in the task vs. the baseline period (bar graph in Fig. 4D). In a final step, the algorithm refines its estimates of task-related locations by determining which locations had a statistically significant increase in the number of detected onsets in the task period as compared to the baseline period. This information is used to further refine the previously identified network, resulting in a final definition of the task-related network (arrow to B from subpanel in D).