Technology & ideas

Behavioral research, human factors, and AI at work: practice, analysis, and methods to be tested.

All 22 essays

Grouped by subject · Most recently updated first within each subject

Reframing, knowledge & judgment

  1. Reframing problems & judgmentAnalysis6 min readDo not mistake people's decisions for natural laws of a systemBeyond repeated outcomes lie actors, information, incentives, authority, and feedback. Examining those mechanisms without inventing access to internal motives.Read essay
  2. Reframing problems & judgmentAnalysis6 min readTechnical sophistication still needs a delivery argumentFrom demonstration to sustained use: comparing coverage, validation effort, maintenance, and exit options. Both simple and complex approaches need to justify their costs.Read essay
  3. Core argumentAnalysis8 min readEvery problem is a technical problemBy technical, I mean examining goals, constraints, and relationships, then reorganizing what is available into an approach we can act on and test. How this shapes my research, use of AI, and collaboration.Read essay
  4. Reframing problems & knowledgeAnalysis6 min readAfter the conversation: how knowledge enters a team's workBetween finding a PDF and reusing a judgment lie context, versions, access, and collaboration. Knowledge engineering continues through actual use and correction.Read essay
  5. Reframing problems & judgmentAnalysis6 min readAn ordinary starting point is a condition of the comparisonBefore comparing career or business paths, define starting conditions, denominators, time, and outcomes. Population distributions cannot directly answer an individual's chances.Read essay
  6. Solution EngineeringPractice reflection2 min readWhat is worth keeping after a project?Beyond the report: reasons for choices, mistakes, and enough instructions for the next person to start.Read essay

AI systems & human collaboration

  1. AI systems & human collaborationMethod to be tested6 min readWhat has an agent automated if someone must keep watching it?Background execution is a beginning. Configuration, supervision, recovery, and interruptions must count when evaluating whether automation makes work easier.Read essay
  2. AI systems & human collaborationMethod to be tested6 min readFrom one study to a reusable skill: retaining the professional methodA good answer does not establish a repeatable process. Extracting inputs, search strategy, evidence, judgment, exceptions, and acceptance from research work.Read essay
  3. AI systems & human collaborationAnalysis6 min readWhen more work becomes programmable, what changes for people?From selecting software to specifying a task: why goals, context, acceptance, and authority matter when agents can connect tools on demand.Read essay
  4. Agentic AIPractice reflection2 min readWhat has to happen before AI can analyze the data?Participant identity, clocks, missing segments, and quality checks need attention before a model begins explaining results.Read essay
  5. Agentic AIPractice reflection2 min readWhat I need to trace behind an AI conclusionSix kinds of checks in the PsyPhiClaw design, from wrong inputs to overstatement, and the validation still needed.Read essay
  6. Agentic AIPractice reflection2 min readWhy my two agents keep separate memoryFrequent briefings and business analysis need different context, with separate decisions about access to key records.Read essay

Behavior, measurement & interpretation

  1. Behavior, measurement & interpretationAnalysis6 min readWhat survives when we turn behavior into labels?An action can be recognized correctly and interpreted incorrectly. What coding schemes preserve, what they exclude, and why behavioral analysis needs a route back to context.Read essay
  2. Behavior, measurement & interpretationAnalysis6 min readAfter recognition: moving from labels to behavioral structureIdentical action counts can describe different processes. Choosing observation units, Markov chains, hidden states, and longer sequences around the question being asked.Read essay
  3. Behavior, measurement & interpretationAnalysis6 min readPosture, expression, and speech: separating observation from inferenceAnother signal is not automatically independent evidence. How multimodal analysis can preserve competing interpretations, disagreement, and missing information.Read essay
  4. Behavioral intelligencePractice reflection2 min readWhen is an extra measurement worth collecting?How I assigned EEG, gaze, and observation different jobs in a packaging study, and how I decide whether another signal helps.Read essay
  5. Working pathPractice reflection2 min readWhat four years of teaching left me withFrom the classroom to field research, I learned to ask where someone struggles to act on an explanation.Read essay
  6. Behavioral intelligencePractice reflection2 min readIs a pause actually a usability problem?A lesson from the six-city finance-app study: preserve the action and its context before deciding what deserves a change.Read essay
  7. Behavioral intelligencePractice reflection2 min readWhat if the first pass misses an event?A tradeoff in the BehaviorLens prototype: intermediate steps aid review, but later passes over candidates cannot recover omitted events.Read essay

Simulation, experience & validation

  1. Simulation, experience & validationAnalysis6 min readAfter moving in: why experience needs different time scalesFirst use, repeated routines, and long-term adaptation ask different questions. Connecting task design and experience storyboards to decisions that can be tested.Read essay
  2. Simulation, experience & validationMethod to be tested6 min readVirtual users should be able to fail a reality checkMoving from plausible responses to testable expectations: defining validation scope, separating calibration data, comparing baselines, and learning from model failure.Read essay
  3. Human FactorsPractice reflection2 min readWhat is this stage of testing trying to answer?How I distinguish early learning from later validation when planning medical-device usability work.Read essay
Want a sequence? Explore four reading paths

Follow the question further

Four connected reading paths.

Each essay stands on its own. Read in sequence to see how one judgment leads to the next question.

From actions to understanding

Examine what labels retain, how events form a process, and what different evidence can support.

  1. What survives when we turn behavior into labels?
  2. After recognition: moving from labels to behavioral structure
  3. Posture, expression, and speech: separating observation from inference
  4. What if the first pass misses an event?

From simulation to reality checks

Make testable expectations, examine tasks across time, and connect validation to development.

  1. Virtual users should be able to fail a reality check
  2. After moving in: why experience needs different time scales
  3. What is this stage of testing trying to answer?

From one execution to sustained work

Define executable tasks, account for human effort, and preserve methods and knowledge for reuse.

  1. When more work becomes programmable, what changes for people?
  2. What has an agent automated if someone must keep watching it?
  3. From one study to a reusable skill: retaining the professional method
  4. After the conversation: how knowledge enters a team's work

From conclusions to their conditions

Ask whom a comparison describes, who can change the rules, and what makes delivery feasible.

  1. Every problem is a technical problem
  2. An ordinary starting point is a condition of the comparison
  3. Do not mistake people's decisions for natural laws of a system
  4. Technical sophistication still needs a delivery argument