Connectionist Priming for Interface Habituation Research

Description

Uses priming paradigms to measure how learned interface conventions pre-activate expectations, predicting errors when a new design partially matches a familiar pattern.

Connectionist models explain priming as residual or spreading activation in a network of associated units: exposure to part of a learned pattern pre-activates the rest. For interfaces, years of exposure to conventions (a red button meaning danger, a gear meaning settings, bottom-right meaning proceed) create strong patterns. When a new design partially matches one, the remaining expectations are activated automatically, which can speed use when consistent or cause capture errors when not. Priming paradigms quantify this: participants are briefly exposed to conventional or modified elements, then respond to a target, and facilitation or interference in response time and error rate measures the strength of the habit. The method predicts where deviating from a convention will be costly and where a new convention can be learned.

Attributes

Loss type
Extinct
Effort
Weeks

Steps

  1. Design stimuli varying the degree of prototype match (0%, 25%, 50%, 75%, 100% of features matching a learned interface pattern). Measure:
  2. response time as a function of match level,
  3. error type when match is partial (do errors reflect the prototype or a novel response?). Errors that match the prototype even at 50% match = strong connectionist priming. Use to predict error patterns after design changes.

Sources

  • Parallel Distributed Processing — David E. Rumelhart
  • McClelland & Rumelhart, An interactive activation model of context effects in letter perception (Psychological Review, 1981)

Related talks

Also known as

  • Priming studies of interface conventions
  • Interactive activation analysis of conventions

Categories

  • User Research
  • Cognitive Psychology & Perception