A challenging problem for current cognitive science would be to take an innovative display, such in the following figure, and generate zero-parameter predictions as to how long it would take an average human to find and extract the answer to a given query.
The eye saccades many times per second. Is it saccading to and fixating on the important information or is it being distracted by visually salient nodes and edges that have low semantic relevance for the query being pursued? When the analyst finds a few anomalous nodes, she imposes semantic meaning on them that they did not have for her before she found them; for this scenario, she has identified them as potential indicators of the anomalous pattern of interaction that she is seeking to uncover. Does the visual saliency of those nodes facilitate or inhibit her search for others? Are the important nodes visually similar to each other and dissimilar to others? If not, can the operator somehow mark the nodes so that she can create subsets of nodes which are visually similar to each other but visually dissimilar from the rest; thereby, creating an onscreen record of her search and of her currently favored hypotheses? Once discovered are these nodes so similar to others that they get “lost” every time she glances away from them to search for others? Or do they get “lost” every time she clicks on another node because the pattern of saliency has changed (as per the left vs right versions of the figure)?