Solution latency per trial
Time from trial start to successful goal entry for each trial and difficulty level.
CSVJSON
← Capabilities
Puzzle Box Problem-Solving
Automated problem-solving assessment in the puzzle box
Quantify latency to solve, strategy type, and learning curves across sequential trials — replacing continuous manual observation with frame-accurate automated scoring.
5
Difficulty levels supported
4
Strategy types classified
33ms
Latency timing precision
100%
Automated scoring pipeline
The problem
Puzzle box scoring demands continuous undivided attention
The puzzle box test requires an observer to watch every second of every trial, coding strategy type and timing solution latency simultaneously. Strategy coding is subjective, and multi-trial experiments with increasing difficulty levels multiply the scoring burden.
- Continuous observation for 5-10 minutes per trial across 7+ difficulty levels per animal
- Strategy coding (e.g., direct entry vs. exploration vs. perseveration) varies across scorers
- Learning curves require consistent scoring across all trials — observer fatigue degrades late-trial accuracy
The solution
Event detection and strategy classification from video
ConductVision detects key puzzle box events — door manipulation, entry into goal zone, orientation toward barrier — and classifies solution strategy from the behavioral sequence. Multi-trial learning curves are computed automatically.
- Frame-accurate latency from trial start to successful entry
- Automated strategy classification: direct, exploratory, perseverative, and novel
- Multi-trial learning curves with strategy transitions across difficulty levels
Endpoints
Problem-solving dependent variables
Strategy classification log
Strategy type assigned to each trial with confidence score and supporting behavioral features.
CSV
Learning curve summary
Latency and strategy progression across trials and difficulty levels, with regression fit for learning rate quantification.
CSVPNG
Applications
Cognitive paradigms
Rule-shift and reversal learning
Puzzle box difficulty increases require strategy shifts. Automated scoring detects perseverative errors and successful rule adaptation.
Measures
- Perseverative error count
- Trials to criterion
- Strategy shift latency
Executive function decline
Puzzle box performance declines with age and in AD/PD models. Automated scoring enables longitudinal tracking with consistent criteria.
Measures
- Maximum difficulty reached
- Learning rate by age
- Strategy complexity score
Environmental enrichment effects
Enriched-housing animals solve puzzle box tasks faster and use more sophisticated strategies — quantified automatically.
Measures
- Enriched vs. standard latency
- Strategy type distribution
- Learning rate comparison
Cognitive enhancer screening
Puzzle box is sensitive to cognitive enhancers and impairing drugs. Automated scoring enables high-throughput dose-response.
Measures
- Dose-dependent latency
- Strategy sophistication by dose
- Error rate
Compared to typical systems
How ConductVision differs
| Feature | ConductVision | Typical systems |
|---|---|---|
| Strategy classification | Automated, 4 types | Manual subjective coding |
| Latency precision | Frame-accurate (33 ms) | Manual stopwatch |
| Multi-trial learning curves | Automatic with regression fit | Manual spreadsheet calculation |
| Difficulty level tracking | Integrated protocol management | Manual trial-by-trial notes |
| Batch processing | Unlimited sessions overnight | Real-time manual observation |
Related capabilities
Behavior Classification
ML-powered frame-level behavioral classification with temporal smoothing and custom classifier training.
Custom Model Training
Train and validate custom behavior classifiers on lab-specific annotated data.
Precision Tracking
High-resolution 30 fps tracking that captures sub-second behavioral events conventional systems miss.
Automated cognitive assessment from your puzzle box recordings
Upload puzzle box videos and get strategy-classified learning curves without manual scoring.
