John Vervaeke identifies two modes of thinking; “Gestalt” vs “Feature”. You can understand them as modes of thought where one of the two processes involved in Predictive Processing is emphasized over the other – the top-down process or the bottom-up process. Both modes are necessary, but both can be harmful if used in exclusion of the other. His argument, and mine, is that you need to balance, and dynamically shift between both, to effectively perceive and conceptualize the world and be “rational”.
Gestalt Mode is a top-down approach where your mind finds (and invents) patterns to bring meaning to complexity. You see a “whole”, and may lose sight of the individual features. As in you see a forest, not the trees. Appropriate usage of Gestalt Mode helps you to frame problems in terms of concepts, models, and systems, which are more easily managed. Overdoing the Gestalt Mode leads to conspiracy theories, religion, and spiritual experiences, as Scott Alexander points out in his essay on “Mysticism and Pattern Matching“. (Worth the whole read)
Once the pattern-matching faculty is way way way overactive, it (spuriously) hallucinates a top-down abstract pattern in the whole universe. This is the experience that mystics describe as “everything is connected” or “all is one”, or “everything makes sense” or “everything in the universe is good and there for a purpose”. The discovery of a beautiful all-encompassing pattern in the universe is understandably associated with “seeing God”.
Feature Mode is a bottom-up approach where the one becomes many. Where you break the “whole” down into its individual components such that you may not even see the “whole” anymore. As in, the forest is more real than the ecosystem. The tree more real than the forest. The atoms more real than the tree. Individual elements in a system become most real, and larger patterns lose meaning and significance. Appropriate usage of Feature Mode is basically what science is – breaking down complex problems into increasingly smaller subcomponents which are more tractable and easy to manipulate. Vervaeke calls the scientific revolution essentially a Feature approach to the world. Another word for this process is reductionism.
The downside of Feature Mode, according to Vervaeke, is the meaning crisis. It’s a mode of thinking which kills meaning, as all concepts become illusions and mere social constructs. And you lose God because you lose sight of the gestalt of the entirety of “creation”.
Someone once told me something that really embodies this way of thinking.
The truth is I’m a collection of particles hurtling through space on another larger collection of particles. I’m an eddy in the universe that will dissipate as quickly as I’ve appeared.
The most extreme version of Feature Mode happens in meditation, and can even be harmful, as pointed out by Holly Elmore in an essay where she explains how meditation has permanently(?) harmed her ability to recognize concepts (among other harms).
One of the general things that mindfulness meditation aims to do is teach the practitioner to perceive sense data more directly and less filtered through preconceived ideas of what it is we’re sensing. It seeks to show us that concepts are an illusion, everything from thinking you see a “table” instead of a composition of light and shadow all the way up to our own self-concepts. The biggest harm of reducing the tendency to pre-filter input through concepts is the processing time that it takes to bind all the shapes or sounds or ideas I’m hearing into something my brain can use.
To be functional in the world, to be rational, you need to have both modes. You neither want to always collapse all meaningful differences into a unitary whole, or be unable to collapse data into larger and more meaningful concepts. You need to be able to move up and down the ladder of abstraction so that you can realize what is relevant and be able to frame problems into a meaningful whole.
These two modes of thought need to feed off each other in a dynamic and adaptive way in order to function at all. After all, if you cannot see the pattern, how can you know what data points are part of the pattern? But if you don’t know what data points are relevant, how can you see the pattern? The circularity here must be solved in a dynamic way where there is a constant scaling, back and forth, between Feature and Gestalt modes of thinking.
While I am unsure whether the meditative and contemplative practices that Vervaeke advocates for will enhance ones ability for Gestalt and Feature modes of thinking respectively, I think Vervaeke is onto something here. Rationality, and all problem solving, seems to require both modes to be feeding off each other. It requires building up possible gestalts (frames, patterns, models), and also seeing the features from which the gestalts are made in order to validate or falsify the gestalt. Indeed, I think this is what the work of Gary Klein shows; experts are just people who have more patterns and are better at pattern recognition.
This is how I currently think about rationality, and I hope to write a piece soon that explains all this in more detail.