computer-science linguistics mathematics medicine-and-health psychology visual-arts

​

Figure ground

​

Description

Figure-ground names the perceptual operation by which a continuous visual (or, by extension, informational) field is segmented into an attended figure and a receding ground. Rubin’s 1915 demonstration of the vase / two-faces image is the textbook case: the same image can be read either way, but never as both at once — perception commits to one assignment and pushes the alternative out of awareness until it is deliberately re-segmented. The concept’s load-bearing claim is that this segmentation is constructed by the perceiver, not given by the stimulus. The same input admits multiple valid figure-ground assignments; which one wins depends on the segmentation rule the observer’s system applies (closure, convexity, symmetry, motion, attention, prior knowledge, top-down framing). Once an assignment is made, the figure inherits properties — boundedness, shape, name, salience — that the ground does not, and the ground is perceived as continuing behind the figure without independent shape. The diagnostic question — “who decided which region is the figure?” — surfaces the contingency. In static images the assignment can be forced by design cues (Kanizsa triangles, depth cues, motion). In informational fields, the assignment is forced by attention budget, framing, and the conventions of the observer’s discipline. Reframing — choosing differently which region of a problem to read as figure — is one of the highest-leverage cognitive operations precisely because most problems’ ground contains the unattended structure that resolves them.

​

Triggers

User-initiated: User is wrestling with how to frame a problem, or feels stuck because “the answer must be somewhere I’m not looking.” Vocabulary cues: “I keep focusing on X but,” “what am I missing?”, “reframe,” “signal vs noise,” “what stands out,” “pop out,” “I’m not seeing it,” “step back.” Agent-initiated: Agent notices the user has committed early to a figure-ground assignment and the structural answer probably lives in the unattended ground. Candidate inference: “the question isn’t about the figure you’ve chosen — it’s about what you’ve parked as ground; what happens if you re-segment?” Situation-shape signals: Bug-hunts that have plateaued. Strategic decisions framed as “do A or B?” where the productive move is “neither — re-segment what counts as the decision.” Design reviews where the surface element gets all the attention and the structural choice is in what’s not on the surface. Political / cultural disagreements where the parties agree on facts but disagree on which facts are figure.

​

Exclusions

• Truly uniform fields with no segmentable structure — a field with no cues for figure-ground assignment (pure white noise, completely uniform color) doesn’t admit the concept. The concept presupposes segmentable structure; “no segmentation possible” is a different situation than “competing valid segmentations.”
• Stimulus-driven mandatory assignments — sometimes the segmentation cues are so strong that assignment is forced and no plausible alternative exists. A small bright object on a uniform dark field is figure; reading the dark field as figure is not perceptually achievable. The concept’s interesting work is in the plastic zone where the observer has a choice; the rigid-assignment case sits outside that zone.
• One-region scenes — if the entire field is figure (a portrait at maximum zoom, an isolated object on a featureless background), the figure-ground distinction collapses into figure-only and loses analytic traction. The concept requires the contrast between figure and ground to do its work.
• Pure quantitative-ranking contexts — when the question is “rank these items by metric X,” not “which of these is the foreground,” the figure-ground concept misframes the task. Ranking presupposes all items are figures; figure-ground asks the prior segmentation question.

​

Structure

​

Relationships

• negative-space — the design discipline that deliberately exploits figure-ground assignment. Figure-ground is the perceptual primitive; negative-space is the move that pre-decides which assignment the perceiver will make. Reading both: negative-space is a design choice; figure-ground is the cognitive substrate that makes the choice possible.
• schema-anomaly — the semantic-level figure-ground event. The anomaly becomes a figure precisely by violating the ground-assumptions of the schema. Skilled diagnosticians train their figure-ground re-segmentation reflex.
• surface — designing a surface is incomplete without designing the figure-ground assignment the surface will provoke in the perceiver.
• load-bearing — load-bearing is the structural-significance question; figure-ground is the prior perceptual-segmentation question that decides at what granularity the load-bearing question gets asked.
• red-herring — exploits the plasticity of figure-ground assignment. The misdirection promotes an irrelevant element to figure-status; the load-bearing element stays in ground.
• reframe — reframe IS a deliberate figure-ground re-assignment at the semantic level. Most reframes come down to “stop reading X as figure; read Y as figure instead.”

​

Examples