You will not find that folder inside the computer
Donald Hoffman compares perception to a computer folder icon: a useful symbol that need not resemble the machinery behind it. This analogy introduces his interface theory. The mathematical models still need to establish how well it describes perception in living organisms.
In his analogy, an interface helps precisely by concealing unnecessary detail. You do not need to understand a circuit to save a picture. Everyday perception comes with a complication: there is no button that switches from the interface to a view outside it. Our eyes are part of the system we are trying to investigate.
Evolution does not reward accuracy alone
In a 2015 paper, Hoffman, Manish Singh and Chetan Prakash compared perceptual strategies in evolutionary games. Simulated organisms competed in worlds defined by the researchers. Under the conditions studied, strategies tuned to useful outcomes could defeat strategies tracking the world's structure. How states of the world translate into fitness matters to the result.
Consider an illustration of our own. A thermometer supplies a precise number; a warning light supplies only 'too hot'. For someone who must withdraw a hand immediately, the warning may be enough. This does not prove Hoffman's theory. It illustrates why useful information and detailed information are different achievements, even when they concern the same situation.
Where the mathematical result stops
The later Fitness Beats Truth work, published in 2021, quantified the advantage of fitness oriented strategies within the researchers' mathematical framework. Hoffman regards it as support for a species specific perceptual interface. This is an argument whose assumptions and models can be examined, rather than settled by insisting that the world feels real.
A competition between defined strategies is not a photograph of reality behind perception. Applying it to human vision requires evaluating how well its assumptions represent living systems. The model gives us a reason to examine confidence in our senses. It does not grant its author special access to the world those senses supposedly conceal.
The further step toward consciousness
In their 2014 paper Objects of consciousness, Hoffman and Prakash began with agents described through experiences, decisions and actions, and developed mathematical accounts of their interactions. They investigated how descriptions of objects and spacetime might emerge. This belongs to Hoffman's wider project of treating consciousness as fundamental.
The claim that senses filter information and the claim that consciousness underlies the physical world require different tests. Evolutionary games examine perceptual strategies under defined conditions. The conscious agent proposal must separately produce predictions that can be compared with measurements.
The approaching truck is still dangerous
In his TED talk, Hoffman addresses the obvious objection: if objects are an interface, why not ignore an approaching truck? His answer distinguishes taking perceptions seriously from taking them as literal descriptions of fundamental reality. A warning can save your life without resembling the machinery behind the danger.
In the accompanying talk, Hoffman explains the computer screen analogy and the conclusions he draws about perception. The useful symbol in the example and the wider claim about perception and reality need to be considered separately.
Hear the researcher
Donald Hoffman at TED2015: Do we see reality as it is? In English.