Why good slides are almost empty

The Miller number everyone cites wrong

In 1956 George Miller wrote an article titled "The Magical Number Seven, Plus or Minus Two". It has become one of the most cited, and widely misread, papers in cognitive psychology. Miller wrote about short-term memory capacity, and the number 7±2 referred to chunks, or meaningful groupings of information, in immediate recall tasks.

Later research has made clear that number was too generous. Nelson Cowan summarized in a 2001 review that the real capacity in short-term memory is about four items, in practice three to five.

Novel material has to fit within this small capacity regardless of the presenter's skill or the audience's interest. Eleven bullets competing with a spoken explanation exceed it quickly. Experts can group familiar information into larger chunks, which expands their effective capacity; most presentations also include material that is new to at least part of the audience.

Sweller: three types of cognitive load

John Sweller laid the foundation in 1988 for what later became known as cognitive load theory. Together with van Merriënboer and Paas he formalized in 1998 a three-way split of the load that has stood the test of time:

Keep the difficulty that belongs to the subject and remove the work created by its presentation. Every unnecessary label, decoration, or repeated sentence leaves less capacity for understanding the content.

The redundancy effect: why reading slides aloud makes it worse

Richard Mayer at UC Santa Barbara has systematically studied what works in multimedia learning. One of his counterintuitive findings is the redundancy effect, documented by Kalyuga, Chandler and Sweller (1999) and replicated since: when the same information is presented both as on-screen text and as spoken narration, audiences learn less than if they only got the spoken version.

Written and spoken language draw on the same verbal channel. When both deliver the same sentences at once, the audience switches between reading and listening, adding effort without adding information.

The exceptions matter: technical terms, new names, key numbers that are hard to catch by ear benefit from being visibly written. But whole sentences that repeat what the presenter is saying hurt comprehension. If it is on the screen, the voice should add something else. If the voice covers it, the slides should be almost empty.

The coherence principle: remove irrelevant detail

Across multiple studies, Mayer tested versions that added interesting but irrelevant details, decorative images, or background music. Participants given the simpler material learned more.

Participants who got the simpler version remembered more and transferred better to new situations. The coherence principle: remove anything that does not contribute. It is one of the most consistent recommendations in multimedia-learning research.

This is also why decorative stock images rarely help. "Image of team high-fiving" next to a strategy slide says nothing about the strategy, but pulls attention, and each gaze shift costs attention that was going to the content.

Why bullets suit documents better than live talks

Hierarchical bullets come from written briefings and memos, particularly the consulting tradition Barbara Minto formalized in The Pyramid Principle. In that structure, each conclusion is supported by a set of sub-arguments that the reader can inspect at their own pace.

These forms were developed for written material meant to be read alone, not for material meant to accompany a voice in a room. When PowerPoint carried the bullet structure forward from 1987 onward, it inherited a rhetorical form originally built for a different use. Bullets work well in written briefings because the reader controls the pace. In a live presentation they compete with the voice for the same verbal channel, and some of both gets lost, as Mayer's redundancy research shows.

Bullets remain useful when the audience needs to scan a written list. During a live talk, use them selectively and avoid reading them word for word. Text and voice should contribute different information or direct attention to different parts of the same idea.

The Tufte critique: what PowerPoint really is

Edward Tufte, the statistician and visualization theorist, wrote the 2003 essay "The Cognitive Style of PowerPoint". He argued that PowerPoint's default templates push presenters toward hierarchical bullet points that flatten complex relationships: precise enough to look analytical, yet too shallow to carry the analysis.

His most unsettling example was the Columbia space shuttle. The Boeing engineers analyzing the foam-strike damage presented their assessment in PowerPoint slides, which passed through several levels of review on the way to NASA management. Tufte's reading of those slides was that the format itself demoted the safety-critical uncertainty: caveats and worst-case scenarios ended up several layers deep in sub-bullets, stylistically equated with administrative details. The information was there. The format made it hard to see.

The point generalizes: when you force complex content into a template that sorts everything into a three-point hierarchy, the structure hides the differences that actually matter. Tufte recommended written notes for analysis and the use of slides only when they contribute something visual, such as a chart, an image, or a table.

What works in practice

The limits of this advice

The cognitive-load findings are well-replicated, but several contexts break the pattern. For an audience with extensive prior knowledge, what looks like overflow is often quickly recognizable. This is the expertise reversal effect that Sweller and Kalyuga documented in later work: what helps novices can slow experts, and vice versa.

Technical reference documents and data-heavy analyses are also exceptions. When the audience is scrutinizing details rather than following an argument, dense information can be the right call. Treat projected reference material as a different genre with its own design requirements.

The same applies when the deck is the document. A board deck read before a meeting, a pitch sent over email, a silent-reading briefing of the kind Amazon uses: in all of these, text and voice don't run at the same time. The reader sets the pace. It still matters that the content doesn't crowd itself out, but the tradeoff between voice and text falls away. The advice in this article applies to a presenter speaking to a live audience. Documents designed for silent reading need a different balance of text and visuals.

Sparse slides can feel unfinished because they do not double as speaker notes. Their job is to direct the audience's attention while the presenter supplies the explanation. That division of work makes the deck easier to follow, even when it gives the presenter less text to stand behind.

A dense slide can reassure the presenter because every talking point is visible. For the audience, processing that layout competes directly with understanding the content. Decide what belongs in the visual and what belongs in the spoken explanation before the session, so the audience does not have to resolve the competition in real time.

Common questions

Sources

• Miller, G. A. (1956). The Magical Number Seven, Plus or Minus Two. Psychological Review, 63(2), 81-97.
• Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences, 24(1), 87-114.
• Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257-285.
• Kalyuga, S., Chandler, P. & Sweller, J. (1999). Managing split-attention and redundancy in multimedia instruction. Applied Cognitive Psychology, 13(4), 351–371. (Primary source for the redundancy effect; see also Mayer’s synthesis in The Cambridge Handbook of Multimedia Learning, 2nd ed., 2014.)
• Minto, B. (1987). The Pyramid Principle: Logic in Writing and Thinking. Pitman. (The principle was developed while the author worked at McKinsey in the 1960s and 70s.)
• Sweller, J., van Merriënboer, J. J. G. & Paas, F. G. W. C. (1998). Cognitive Architecture and Instructional Design. Educational Psychology Review, 10(3), 251–296.
• Tufte, E. R. (2003). The Cognitive Style of PowerPoint: Pitching Out Corrupts Within. Graphics Press.
• Kalyuga, S., Ayres, P., Chandler, P. & Sweller, J. (2003). The expertise reversal effect. Educational Psychologist, 38(1), 23-31.