Search Handbook Cards

Decline in working memory capacity and updating speed

The number of items working memory can hold at once is limited

How many options you can speak is capped by working memory

Without rehearsal, working memory content decays naturally within roughly ten to twenty seconds

Presenting the same information through both visual and verbal channels can raise working memory's effective capacity

Retention is brief and vulnerable to interference

The classic capacity estimate is closer to four chunks than the earlier figure of seven

The visuospatial sketchpad holds visual and spatial information

The three load types draw on one shared, limited processing capacity

The two subsystems draw on relatively independent resources, so simultaneous information in different channels doesn't interfere

The longer feedback takes to arrive, the less likely anyone is to catch their own mistake

The phonological loop holds verbal and numeric information

Speech conveys meaning well but structure poorly

Requiring users to remember content across pages is itself a design flaw

The cap is readable width, not memory span

Within a one-second delay, the user's train of thought stays intact

Capacity varies with item complexity, so simple and complex items can't be compared on one number

Post-completion error at the tail of a sequence

Facts from other pages must remain visible here

A dense but well-structured interface can beat a sparse but chaotic one

Capacity limits and time-based decay are two independent mechanisms, and easing one doesn't substitute for the other

A measurable resumption lag exists before performance returns to pre-interruption levels

Action sequences can be automatized into unconscious execution

Grouping a dozen digits into a familiar pattern turns them into just a few items to hold

Domain experts tolerate higher information density

Breakpoints come in coarse and fine grain, and the subtask boundary is only one coarser level

Users should not have to remember information across screens

Reduce the number of choices shown at once

Moderate time pressure raises alertness and engagement, but excessive pressure crowds out processing capacity

Walking, driving, or minding a child leaves little spare attention for the interface

A number that changes mid-glance interrupts exactly the moment a reader was trying to lock onto it

Cognitive distraction is attention leaving the driving task

Rehearsal extends retention, but draws on the same limited processing resource

Interruption cost rises with how deep the task has gone

The more information an interruption shares with the primary task, the more reconstruction work resumption requires

Long lists are extremely costly in the auditory channel

Sparse layouts levy a paging and scrolling tax

An expert can chunk new information into a few packages; a novice has no packaging to use

External representation moves computation and comparison onto the interface, cutting mental arithmetic and tracking load

Mode ambiguity needs its own lamp or legend, not reliance on remembering the current mode

Introduce one mechanic at a time

Selecting points in one view should light up the same data wherever else it appears on the dashboard

The current task goal determines which features get prioritized in search

Forgetting is fastest right after learning

Reduce short-term memory load

The intrinsic-extraneous-germane split rests on schema theory, where germane load builds the schema

Loss of lens accommodation and near-vision difficulty

A dashboard should read in the same order as the decision it supports: check, locate, then decide

Seeing the actual value without its target next to it hides how far off the process really is

A full screen briefly registers whole in sensory memory, then almost all of it is gone