Layering multiple encodings cancels the parallel advantage

What it is

Conjunction-search cost arises when no single feature defines a target and the viewer must find a combination such as “red and triangular.” If red circles and blue triangles are distractors, neither feature alone identifies the target, so search usually relies more heavily on attentional guidance. “Cancels” is a design warning, not a claim that every multi-feature display becomes strictly serial. Redundantly encoding one variable also differs from defining a target by a conjunction.

Why it happens

Multiple features can jointly guide attention, but speed depends on whether a feature or combination is diagnostic within the current distractor structure. Heterogeneous distractors, crowding, target uncertainty, and similar feature values weaken guidance. If color encodes state and shape encodes category, locating one category-by-state requires binding dimensions. If both color and shape encode the same selection, they may provide redundant gain instead.

Studying it

At matched set sizes, compare single-feature, conjunction, and redundant targets with both target-present and target-absent trials. Measure response-time slope, errors, eye movements, and learning while matching the individual discriminability of features. Vary distractor combinations and target priors to separate feature effects from probability and layout grouping.

Where it stops holding

Familiar targets, sparse layouts, strong semantics, and staged filtering can guide conjunction search. Complex encoding may serve deliberate lookup after browsing rather than instantaneous discovery. Redundant encoding of one variable can improve robustness; the concern is requiring a logical AND across independent dimensions, not every chart that uses several channels.

Applying it

• Write the target predicate. If it requires several dimensions, add filtering, facets, sorting, or search to reduce the candidate set.
• Give urgent targets one cue sufficient for location, with labels or shape as accessible redundancy rather than another independent condition.
• Compare single-feature and conjunction versions across target presence, density, and expertise, using search-time slope and errors to revise the design.
• State whether channels encode the same or different variables, and provide equivalent combined filters and result counts nonvisually.

Related

• Same group: U1.12.1 A single salient difference can be found in parallel · U1.12.3 Highlight only the class currently in focus
• Adjacent: U1.10.1 Encoding one variable on two channels improves judgment robustness · U1.06.4 Lightness and saturation changes are read as one strength dimension
• Search terms: conjunction search · guided search · redundant target