http://www.idiap.ch/uist2006/ Using a Low-Cost Electroencephalograph for Task Classification in HCI Research en Tue, 17 Oct 2006 00:00:00 +0200 no Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research Desney S. Tan cane UnMa U IST 2006, Montreux Switz, iand Using a Low-Cost Electroencephalograph for Task Classification in HCI Research Desney S. Tan cane UnMa U IST 2006, Montreux Switz, iand 00:00:07 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 On page 81 of the pnnt proceedings On page 81 of the pnnt proceedings 00:00:24 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research Desney S. Tan university U IST 2006, Montreux Switz, tand Using a Low-Cost Electroencephalograph for Task Classification in HCI Research Desney S. Tan university U IST 2006, Montreux Switz, tand 00:00:08 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 NATIONAL G C NY limes M Oc 16. 2005 NATIONAL G C NY limes M Oc 16. 2005 00:00:35 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Brain-Computer Interfaces BCI A direct technological interface between a brain and computer not requlnng any motor output from the user Example ConferencesJoumals with BCI interests Neural Inforrnatlon Processing Systems NIPS Transactions on Biomedical Englneenng Transactlons on Neural Systems and Rehabiirtatlon Engineenng Brain-Computer Interfaces BCI A direct technological interface between a brain and computer not requlnng any motor output from the user Example ConferencesJoumals with BCI interests Neural Inforrnatlon Processing Systems NIPS Transactions on Biomedical Englneenng Transactlons on Neural Systems and Rehabiirtatlon Engineenng 00:00:34 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Why is this relevant to UIST or HCI BCI research traditionally focuses on exploratory neuroscience and rehabilitation engineering. Brain sensing could prowde valuable data about engagement D cognitive work load satisfaction Potentially helpful to Context Sensitive or Evaluation Systems Why is this relevant to UIST or HCI BCI research traditionally focuses on exploratory neuroscience and rehabilitation engineering. Brain sensing could prowde valuable data about engagement D cognitive work load satisfaction Potentially helpful to Context Sensitive or Evaluation Systems 00:00:50 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Values of BCI Values of HCI Values of BCI Values of HCI 00:00:10 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Values of BCI Values of HCI To use any means necessary to demonstrate that brain-computer interaction is passible. Values of BCI Values of HCI To use any means necessary to demonstrate that brain-computer interaction is passible. 00:00:20 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Values of BCI Values of HCI To use any means necessary to demonstrate that brain-computer interaction is possible. useempnlem eosmg use highly invasive surgical require hours or ctays of remove data from poor Human-Computer Interactzon instztute Values of BCI Values of HCI To use any means necessary to demonstrate that brain-computer interaction is possible. useempnlem eosmg use highly invasive surgical require hours or ctays of remove data from poor Human-Computer Interactzon instztute 00:00:34 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Values of BCI Values of HCI To use any means necessary to To use reasonable means to achieve a practw.at benefit to demonstrate that brain-computer interaction is passible. use equipment costing use highly invasive surgical require hours or clays of remove data from poor Values of BCI Values of HCI To use any means necessary to To use reasonable means to achieve a practw.at benefit to demonstrate that brain-computer interaction is passible. use equipment costing use highly invasive surgical require hours or clays of remove data from poor 00:00:11 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Values of BCI Values of HCI To use any means necessary to to use reasonable means to achieve a pracbcai benefit to demonstrate that brain-computer interaction is possible. many usem. usefally use equipment cosmg 10OK to I million USD accessil equipment be safe for reoemea aria use highly invase surgical significant user tr aimng remove data from poor Values of BCI Values of HCI To use any means necessary to to use reasonable means to achieve a pracbcai benefit to demonstrate that brain-computer interaction is possible. many usem. usefally use equipment cosmg 10OK to I million USD accessil equipment be safe for reoemea aria use highly invase surgical significant user tr aimng remove data from poor 00:00:48 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Where do we start Where do we start 00:00:04 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Brain SensingImaging Technologies MRI SPECT PET MEG EROSINIR EEG Brain SensingImaging Technologies MRI SPECT PET MEG EROSINIR EEG 00:00:13 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Brain SensingImaging Technologies MRI SPECT Currently Impractical PET for HCI MEG EROSIfNIR EEG Brain SensingImaging Technologies MRI SPECT Currently Impractical PET for HCI MEG EROSIfNIR EEG 00:00:22 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 EEG Electroencephalograph the neumphysiologicat measurement of the electrical actrvity of the brain by recording from electrodes placed on the scalp Measures the voltage difference between two Iocalons on the scalp Only picks up gross, macroscopic, coordinated, and synchronized finng of neurons near the surface of the brain wi perpendcular onentabon to the scalp. thus m of actvrty is hidden EEG Electroencephalograph the neumphysiologicat measurement of the electrical actrvity of the brain by recording from electrodes placed on the scalp Measures the voltage difference between two Iocalons on the scalp Only picks up gross, macroscopic, coordinated, and synchronized finng of neurons near the surface of the brain wi perpendcular onentabon to the scalp. thus m of actvrty is hidden 00:00:39 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 EEG Electroencephalograph the neumphysiologicat measurement of the electrical actrvity of the brain by recording from electrodes placed on the scalp Measures the voltage dfference between two locations on the scalp Only picks up gross, macroscopic, coordinated, and synchronized finng of neurons near bhe surface of the brain wibh perpendCutar onentabon to the scalp. thus majority of actnnty is hidden EEG Electroencephalograph the neumphysiologicat measurement of the electrical actrvity of the brain by recording from electrodes placed on the scalp Measures the voltage dfference between two locations on the scalp Only picks up gross, macroscopic, coordinated, and synchronized finng of neurons near bhe surface of the brain wibh perpendCutar onentabon to the scalp. thus majority of actnnty is hidden 00:00:10 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 EEG Devices M EGt Systems Channels 128-512 Cost 30K USD Cost 100K-250K USD EEG Devices M EGt Systems Channels 128-512 Cost 30K USD Cost 100K-250K USD 00:00:32 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 The Brainmaster Lowest cost FDA approved device Designed far home and small clinical use. Only 1500 USD Specs 2-channels 8-brt at 41JV resotulon 256 samplessec Has yet to be validated for BCI research work. If it works, it lowers the entry bar for BCI research. The Brainmaster Lowest cost FDA approved device Designed far home and small clinical use. Only 1500 USD Specs 2-channels 8-brt at 41JV resotulon 256 samplessec Has yet to be validated for BCI research work. If it works, it lowers the entry bar for BCI research. 00:00:32 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Validating the Device and ourselves 1. Validate the device Can we get useful data from such a low-end device 2. Validate ourselves To explore this space, we must be able to collect our own data. Validating the Device and ourselves 1. Validate the device Can we get useful data from such a low-end device 2. Validate ourselves To explore this space, we must be able to collect our own data. 00:00:14 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Validating the Device and ourselves Keim, Z. A New Mode of Communication Between Man and His Surroundings, IEFE Transactions on Biomedical Englneenng, Vol. 37, Data is available for download Data has not been reproduced in the past 15 years Some computational BCI researchers have just used this data. State of the art does s not a great deal better. Validating the Device and ourselves Keim, Z. A New Mode of Communication Between Man and His Surroundings, IEFE Transactions on Biomedical Englneenng, Vol. 37, Data is available for download Data has not been reproduced in the past 15 years Some computational BCI researchers have just used this data. State of the art does s not a great deal better. 00:00:39 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Reproducing the Keim Data We adapted procedure from Keim to better control potential confounds. Rest Baseline Relaxation and cleanng of mind Math Mental aurTthmet, prompted wrth 7 times 3 8 5 Rotation Mentally rotate an object prompted with peacock Tasks from the original paper were designed to elicit hemispheric differences. Reproducing the Keim Data We adapted procedure from Keim to better control potential confounds. Rest Baseline Relaxation and cleanng of mind Math Mental aurTthmet, prompted wrth 7 times 3 8 5 Rotation Mentally rotate an object prompted with peacock Tasks from the original paper were designed to elicit hemispheric differences. 00:00:50 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Experimental Procedure User is instructed to keep eyes closed, minimize body movement and not to vocalize part of the tasks. For each task, a computer dnven cue given Rest, Math, Rotate Following Math and Rotate, the experimenter says either the math problem or object Experimental Procedure User is instructed to keep eyes closed, minimize body movement and not to vocalize part of the tasks. For each task, a computer dnven cue given Rest, Math, Rotate Following Math and Rotate, the experimenter says either the math problem or object 00:00:34 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Experimental Procedure Ellotk amagn ntmOBl from KNml Experimental Procedure Ellotk amagn ntmOBl from KNml 00:00:10 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Experimental Procedure Eilotk amagn aaa from Kem Human-Computer Interactzon Instrtute Experimental Procedure Eilotk amagn aaa from Kem Human-Computer Interactzon Instrtute 00:00:08 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Experimental Procedure iock uegn from KJem Rest Math Rest Math Math Rot Rest SeSSK Experimental Procedure iock uegn from KJem Rest Math Rest Math Math Rot Rest SeSSK 00:00:05 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Experimental Procedure Rot Math Rest Math Rest Fot Math Rest Math Rest Math Rest Rest Rat Math Rest I Math Rest Math Rest Math Rest Math Rest Math Mmh Rot Rest Math Rest Math Rest 3 sesons per subject Many short tasks prevent con etation wdh EB EEG drift Experimental Procedure Rot Math Rest Math Rest Fot Math Rest Math Rest Math Rest Rest Rat Math Rest I Math Rest Math Rest Math Rest Math Rest Math Mmh Rot Rest Math Rest Math Rest 3 sesons per subject Many short tasks prevent con etation wdh EB EEG drift 00:00:26 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Experimental Procedure Math Rest Math Rest Rat Math Fest Math Rest Rest Rat Math Rest Rot Math Rest Math Rest Math Rest Math It Rest Math Math Rat Rest Math Rot Rest Mattt Rest 29-58 years of age All were cognitweiy and neurologically heatthy All nght handed Human-Computer Interactzon instztute Experimental Procedure Math Rest Math Rest Rat Math Fest Math Rest Rest Rat Math Rest Rot Math Rest Math Rest Math Rest Math It Rest Math Math Rat Rest Math Rot Rest Mattt Rest 29-58 years of age All were cognitweiy and neurologically heatthy All nght handed Human-Computer Interactzon instztute 00:00:11 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 EEG Setup Intemabonal 10-20 EEG electrode placement system Two charmels placed on P3 and P4 wdh both references to Cz Electrodes are held in place using conducte paste. EEG Setup Intemabonal 10-20 EEG electrode placement system Two charmels placed on P3 and P4 wdh both references to Cz Electrodes are held in place using conducte paste. 00:00:48 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research Using a Low-Cost Electroencephalograph for Task Classification in HCI Research 00:00:07 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Data Processing 14 secs Data Processing 14 secs 00:00:44 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Removing time for machine learning Most machine learning algonthms don t handle time senes data very well. 10 seconds Removing time for machine learning Most machine learning algonthms don t handle time senes data very well. 10 seconds 00:00:13 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Removing time for machine learning Divk the 10 seconds into 2 sec windows gzat ovedap by I sec Perform signal processing on each of the 9 windows to get our less feature set Removing time for machine learning Divk the 10 seconds into 2 sec windows gzat ovedap by I sec Perform signal processing on each of the 9 windows to get our less feature set 00:00:10 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Removing time for machine learning Divide the 10 seconds into 2 sec windows that overlap by I sec Perform signal processing on each of the 9 windows to get our time less feature set Removing time for machine learning Divide the 10 seconds into 2 sec windows that overlap by I sec Perform signal processing on each of the 9 windows to get our time less feature set 00:00:05 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Removing time for machine learning Divide the 10 seconds into 2 sec windows that overlap by I sec Perform signal processing on each of the 9 windows to get our time less feature set This provides I 486 windows L per participant Removing time for machine learning Divide the 10 seconds into 2 sec windows that overlap by I sec Perform signal processing on each of the 9 windows to get our time less feature set This provides I 486 windows L per participant 00:00:11 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Signal features for each window Generic signal features such as mean power, peak frequency, peak frequency amplitude, etc. Features frequently used in EEG signal analysis. Signal features for each window Generic signal features such as mean power, peak frequency, peak frequency amplitude, etc. Features frequently used in EEG signal analysis. 00:00:17 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Common EEG Features Common EEG Features 00:00:30 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Feature Processing and Selection The 39 base features from each window are mathematically combined to create 1521 total We used a feature preparation and selection process similar to Fogarty CHI 05 to reduce the number of features 23 features for 3-task classification 486 examples 16.4 features for pair-wise classification 324 examples Feature Processing and Selection The 39 base features from each window are mathematically combined to create 1521 total We used a feature preparation and selection process similar to Fogarty CHI 05 to reduce the number of features 23 features for 3-task classification 486 examples 16.4 features for pair-wise classification 324 examples 00:00:33 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research Using a Low-Cost Electroencephalograph for Task Classification in HCI Research 00:00:23 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research Using a Low-Cost Electroencephalograph for Task Classification in HCI Research 00:00:07 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 00:00:07 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 BE Throwing time back in. We can average over temporally adjacent windows to improve classification accuracy BE Throwing time back in. We can average over temporally adjacent windows to improve classification accuracy 00:00:08 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research Using a Low-Cost Electroencephalograph for Task Classification in HCI Research 00:00:27 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Averaging with Task Transitions Fewer task transibons will yletd better classificabon accuracy. Averaging with Task Transitions Fewer task transibons will yletd better classificabon accuracy. 00:00:12 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Averaging with Task Transitions No ons and averaging over all dala will be the even betler. Averaging with Task Transitions No ons and averaging over all dala will be the even betler. 00:00:15 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research Using a Low-Cost Electroencephalograph for Task Classification in HCI Research 00:00:31 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 So, can we really read minds ID. IDb V. ROt RWv. Rot So, can we really read minds ID. IDb V. ROt RWv. Rot 00:00:06 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research Using a Low-Cost Electroencephalograph for Task Classification in HCI Research 00:00:09 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 CognitiveMotor Fabric Tasks of varying cognitive difficultly are involuntarily coupled with physiological responses, such as minute imperceptible motor activity, pOamer 91 Therefore, it is impossible to completely isolate cognitive activity neurologically intact individuals. CognitiveMotor Fabric Tasks of varying cognitive difficultly are involuntarily coupled with physiological responses, such as minute imperceptible motor activity, pOamer 91 Therefore, it is impossible to completely isolate cognitive activity neurologically intact individuals. 00:00:34 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 CognitiveMotor Fabric Tasks of varying cognitive difficultly are involuntarily coupled with physiological responses, such as minute imperceptible motor activity, gOamer 91 Therefore, it is impossible to completely isolate cognitive activity neurologically intact individuals. Does this matter to neuroscience Yes CognitiveMotor Fabric Tasks of varying cognitive difficultly are involuntarily coupled with physiological responses, such as minute imperceptible motor activity, gOamer 91 Therefore, it is impossible to completely isolate cognitive activity neurologically intact individuals. Does this matter to neuroscience Yes 00:00:08 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 CognitiveMotor Fabric Tasks of varying cognitive difficultly are involuntarily coupled with physiological responses, such as minute imperceptible motor activity, pOamer 91 Therefore, it is impossible to completely isolate cognitive activity neurologically intact individuals. Does this matter to neuroscience Yes Does this matter to HCI Maybe not CognitiveMotor Fabric Tasks of varying cognitive difficultly are involuntarily coupled with physiological responses, such as minute imperceptible motor activity, pOamer 91 Therefore, it is impossible to completely isolate cognitive activity neurologically intact individuals. Does this matter to neuroscience Yes Does this matter to HCI Maybe not 00:00:10 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 CognitiveMotor Fabric If motor arbfacts are reliably correlated wrth different types of tasks or engagement, why not use those to help the classifier. Requinng users to not move is also very impractical. Non-Cognve Artifacts detected by EEG. Eye movement He movement Scalpat GSR Jaw end facial EMG limb mavenents Sensory Response Potentials CognitiveMotor Fabric If motor arbfacts are reliably correlated wrth different types of tasks or engagement, why not use those to help the classifier. Requinng users to not move is also very impractical. Non-Cognve Artifacts detected by EEG. Eye movement He movement Scalpat GSR Jaw end facial EMG limb mavenents Sensory Response Potentials 00:00:37 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Experiment 2 Game Task To explore this idea of using non-cognitive aYdfacts to classify tasks using EEG, we chose a PC-based video game task. Hate, a PC-based first person shooter game produced by Micrasaft Game Studies. Naviga a 313 environment in an effort to shoot opponents using various weapons. Relanvety high aegree of I mouse and keyboard input controls. Experiment 2 Game Task To explore this idea of using non-cognitive aYdfacts to classify tasks using EEG, we chose a PC-based video game task. Hate, a PC-based first person shooter game produced by Micrasaft Game Studies. Naviga a 313 environment in an effort to shoot opponents using various weapons. Relanvety high aegree of I mouse and keyboard input controls. 00:00:42 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Game Tasks Rest basehne rest task relax fixate eyes on cross hairs on center of screen, do not interact with controls. Game do not interact with Solo navigate environment, interact with elements in the scene, and collect ammunibon. Opponent controlled by expert dKt not interact with Play navigate environment and engage opponent controlled by expert. Expert instructed to play at a level just shghtly above skill of pacipant to optimally challenge them. Game Tasks Rest basehne rest task relax fixate eyes on cross hairs on center of screen, do not interact with controls. Game do not interact with Solo navigate environment, interact with elements in the scene, and collect ammunibon. Opponent controlled by expert dKt not interact with Play navigate environment and engage opponent controlled by expert. Expert instructed to play at a level just shghtly above skill of pacipant to optimally challenge them. 00:00:50 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Game Experimental Procedure Setup, design, and procedure was similar to first study. Participants had tutorial and practice time with game controls. 3 tasks repeabad 6 times counterbalanced Tasks were 24 seconds to allow navigation time. Only 2 sessions were run for each participant Same 8 participants from first study were run in this study. Same data preparation and machine learning procedure. Game Experimental Procedure Setup, design, and procedure was similar to first study. Participants had tutorial and practice time with game controls. 3 tasks repeabad 6 times counterbalanced Tasks were 24 seconds to allow navigation time. Only 2 sessions were run for each participant Same 8 participants from first study were run in this study. Same data preparation and machine learning procedure. 00:00:38 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Results Game Tasks Mean CI ficazion Accuracy vs. Averacjng Scenarios Game Tasks V. SOIO Results Game Tasks Mean CI ficazion Accuracy vs. Averacjng Scenarios Game Tasks V. SOIO 00:00:31 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Conclusion This experimental design and data processing procedure can be apphed to a much wider range of apphcabonstasks. Our two expenments were just two examples at different ends of a spectrum. Compelling results can be achieved with low-cost equipment and without significant medical expertise or training. Non-cognmve artifacts tnevrtable tn reattstc computing scenarios can be embraced improve classification power. To make BCI relevant to HCI, we must challenge tradmonal assumptions and creatively work with its limitations. Conclusion This experimental design and data processing procedure can be apphed to a much wider range of apphcabonstasks. Our two expenments were just two examples at different ends of a spectrum. Compelling results can be achieved with low-cost equipment and without significant medical expertise or training. Non-cognmve artifacts tnevrtable tn reattstc computing scenarios can be embraced improve classification power. To make BCI relevant to HCI, we must challenge tradmonal assumptions and creatively work with its limitations. 00:01:06 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Thanks Johnny Chung Lee Desney Tan O OD O O Thanks ID MSR and the VIBE Group for suppormg ths work. Thanks Johnny Chung Lee Desney Tan O OD O O Thanks ID MSR and the VIBE Group for suppormg ths work. 00:05:06 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research Wed, 18 Oct 2006 00:00:00 +0200 Using a Low-Cost Electroencephalograph for Task Classification in HCI Research SCIENCE > 2006 Cross-user Classifier 3 Tasiks mem.qesOd attw Cross-user Classifier 3 Tasiks mem.qesOd attw 00:00:33 no Johnny Lee, Desney Tan, Carnegie Mellon University, Microsoft Research