Computer Assisted Attention Training Guy Mark Lifshitz Supervisor: Clark Verbrugge Co-Supervisor: Amir Raz
The Goal Use computer games to improve the attention span and behaviour of children Future WorkBackgroundGamesDesignDevelopmentResults
Psychology Background:
Types of Attention Alerting Ability to respond to an impending event Orienting Ability to shift focus between stimuli Executive Control over multiple neural areas to resolve conflicts Future WorkBackgroundGamesDesignDevelopmentResults
How Attention Training Works Requires sustained attention Self-regulation/conflict resolution Working Memory Future WorkBackgroundGamesDesignDevelopmentResults
What exists Background:
Existing Games IBM Accessibility Research and The Sackler Institute for Developmental Psychobiology Written in 2001 Future WorkBackgroundGamesDesignDevelopmentResults
Existing Games Joystick based 4 – 7 year olds healthy and ADHD children Future WorkBackgroundGamesDesignDevelopmentResults
Existing Games Joystick Training Future WorkBackgroundGamesDesignDevelopmentResults
Existing Games Conflict Training Future WorkBackgroundGamesDesignDevelopmentResults
My games
Games Created (1/5) Future WorkBackgroundGamesDesignDevelopmentResults
Games Created (1/5) Future WorkBackgroundGamesDesignDevelopmentResults
Games Created (2/5) Future WorkBackgroundGamesDesignDevelopmentResults
Motivating Ideas Primary goal: Train for joystick use Side-benefits: Requires sustained attention Future WorkBackgroundGamesDesignDevelopmentResults
Games Created (3/5) Future WorkBackgroundGamesDesignDevelopmentResults
Motivating Ideas Moving distractions (orienting, executive) Unpredictable location (orienting) Training of sounds (for future games) Future WorkBackgroundGamesDesignDevelopmentResults
Games Created (4/5) Future WorkBackgroundGamesDesignDevelopmentResults
Games Created (4/5) Future WorkBackgroundGamesDesignDevelopmentResults
Motivating Ideas Similar to the Stroop test Child specific considerations: Sound rather than reading Animals rather than colors Green Red Blue Future WorkBackgroundGamesDesignDevelopmentResults
Motivating Ideas Ignore conflicting sound/visual cue (executive) Remember the sound/visual cue (Working memory) Requires automated sound recognition (previous games) Future WorkBackgroundGamesDesignDevelopmentResults
Games Created (5/5) Future WorkBackgroundGamesDesignDevelopmentResults
Games Created (5/5) Future WorkBackgroundGamesDesignDevelopmentResults
Games Created (5/5) Future WorkBackgroundGamesDesignDevelopmentResults
Motivating Ideas Player Specific Feedback (cocktail party effect) Conflict resolution (executive attention) Inconsistent timing (alerting attention) Unpredictable location (orienting) Future WorkBackgroundGamesDesignDevelopmentResults
Development Cycle
Preliminary Research 1. Meet with research assistants 2. Meet with Prof. Raz 3. Read psychology papers 4.Read Child-Computer Interaction (CCI) papers Future WorkBackgroundGamesDesignDevelopmentResults
Development Cycle 1. Talk to RAs 2. Develop prototype 3. Test with Prof. Raz’s children 4. Get feedback from Prof. Raz Repeat Future WorkBackgroundGamesDesignDevelopmentResults
Development Cycle 1. Talk to RAs 2. Develop prototype 3. Test with Prof. Raz’s children 4. Get feedback from Prof. Raz Repeat Future WorkBackgroundGamesDesignDevelopmentResults
Constraints and Considerations
Design Considerations Measurements Repeatability Modifiable Extendible Code reuse Comprehensible to psychologists Future WorkBackgroundGamesDesignDevelopmentResults
Measurements Measure success Measure time Future WorkBackgroundGamesDesignDevelopmentResults
Measurements CSV Level, Reaction Time, Successful 0, 7553, true 1, 1630, true 2, 1416, false 1, 902, true 2, 1242, true Future WorkBackgroundGamesDesignDevelopmentResults
Repeatability Reduce randomized aspects Future WorkBackgroundGamesDesignDevelopmentResults
Repeatability Reduce randomized aspects Drawback: hard to setup Future WorkBackgroundGamesDesignDevelopmentResults
Repeatability Reduce randomized aspects Drawback: hard to setup Unclear how much randomness is acceptable Future WorkBackgroundGamesDesignDevelopmentResults
Modifiable Configuration files # REQUIRED ORDER: # For video : videos must me labeled "0.mpg", "1.mpg" etc.. in order. # Position Of the goal, Right to left, top to bottom, 0 = top left, 7 = bottom right, no middle position 334 # Use same animal as previous round? 0 = use new animal, 1 = use same 110 # Show the goal animal in the middle during gameplay, 0 = no, 1 = yes 110 # Move the goal animal up and down?, 0 = no, 1-9 = slowest to fastest 001 # Move a specific bush, 0-7 = show at that locaion, 9 = show none (Note: this does not neccessarily have to be an occupied bush) 912 # Show a video? 0 = no, 1 = show before, 2 = show after 000 #Other animal positions, 0-7 = position, 9 = don't show #divider to show that these are two different sets ================================================ #Is each obstacle animal moving? 0 = not moving, 1-9 = slowest to fastest Future WorkBackgroundGamesDesignDevelopmentResults
Modifiable Images Sounds Videos Future WorkBackgroundGamesDesignDevelopmentResults
Extendible Plot File Structure Game Structure Future WorkBackgroundGamesDesignDevelopmentResults
Extendible Plot File Structure Game Structure (open ended concept) (can add videos) Future WorkBackgroundGamesDesignDevelopmentResults
Extendible Plot File Structure Game Structure (open ended concept) (can add videos) (profile folders with custom sounds, ideally images in future) Future WorkBackgroundGamesDesignDevelopmentResults
Extendible Plot File Structure Game Structure (open ended concept) (can add videos) (profile folders with custom sounds, ideally images in future) (Games as independent classes) Future WorkBackgroundGamesDesignDevelopmentResults
Code Reuse Abstract classes if possible Abstract game design Future WorkBackgroundGamesDesignDevelopmentResults
Code Reuse Abstract classes if possible Hard to know beforehand how to do these Abstract game design Future WorkBackgroundGamesDesignDevelopmentResults
Final Thoughts
Measuring Success Test Performance after repeated treatment: Conflict Resolution Child ANT (Attention Network Test) Stroop, Simon, Flanker Working Memory N-Back test IQ scores and more Future WorkBackgroundGamesDesignDevelopmentResults
Measuring Success Test Performance after repeated treatment: Not enough time or resources Future WorkBackgroundGamesDesignDevelopmentResults
Measuring Success Preliminary testing on children: Seems Fun Keeps their attention Games intuitive? Future WorkBackgroundGamesDesignDevelopmentResults
Problems Encountered Unclear what input to use Mixed messages Psychology concepts somewhat vague Future WorkBackgroundGamesDesignDevelopmentResults
Problems Encountered Unclear what input to use Mouse, Joystick, Keyboard Mixed messages Psychology concepts somewhat vague Future WorkBackgroundGamesDesignDevelopmentResults
Future Work Create more games Future WorkBackgroundGamesDesignDevelopmentResults
Future Work Run many children with the games I made Compare ANT to first generation of games Test on children with ADHD Measure child game play Head movements Track eyes Future WorkBackgroundGamesDesignDevelopmentResults
References Raz Lab’s games: “Training, maturation, and genetic influences on the development of executive attention”, Posner et al 2005 “Alerting, Orienting, and Executive Attention”, Mezzacappa 2004 “Computer-assisted Attention Training in Schizophrenics”, M. Hermanutz and J. Gestrich 1991 “Training Working Memory in Children With ADHD”, Klingberg et al 2002 “Typologies of attentional networks”, Raz and Buhle 2006 Future WorkBackgroundGamesDesignDevelopmentResults