Color ΔE and why it matters

If your experiment asks participants to discriminate or categorise colours, the perceptual distance between your stimuli is the experiment. Hand-picked hex codes almost always produce uneven distances — blues cluster, greens spread, and luminance drifts with hue. The effect you report is then a blend of category boundaries and raw physical difference.

What ΔE is

ΔE is a single number that expresses the perceived distance between two colours in CIE Lab space. A ΔE of 1 is the smallest difference a trained observer can reliably detect under standardised viewing conditions. ΔE of 5 is clearly distinguishable; ΔE of 10 is »different colours«. The scale is perceptually uniform — a ΔE step of 1 feels the same whether you are moving between two blues or two yellows.

Why hex-picked stimuli go wrong

• Luminance drift. #ff0000 (red) and #00ff00 (green) look equally saturated but differ in luminance by a factor of five. A »red vs green« discrimination task is contaminated by a »dark vs bright« task.
• Uneven ΔE. Adjacent swatches in a hand-picked ramp are rarely equidistant. Participants see some steps as tiny and others as huge; response latencies reflect the step size, not the manipulation you planned.
• Category centre drift. Pick seven »blues« from intuition and you end up with a cluster around true blue plus outliers into teal and violet. Any Winawer-style within-vs-cross-category analysis is then noise.

What SciBLIND does

SciBLIND's stimulus generator fixes L* (luminance) and steps through hue or chroma in uniform ΔE increments. You pick a centre colour, a ΔE step, and a number of swatches; the tool returns a colour-stops JSON, the per-trial ΔE, and a methodology note that cites the generator version so a reviewer can reproduce the ramp.

When you still need a colorimeter

ΔE is calibrated for a reference display and a neutral surround. Field studies in daylight, on phone screens, or under filter-glass conditions need a device calibration pass. SciBLIND flags »uncontrolled viewing« on the export so the reviewer knows the ΔE is a target, not a measured stimulus.