This is an excellent article bringing up the idea of suspension of belief, believed itself to be central to the enjoyment of fiction. However, in computational cognitive neuroscience we have long understood the basic mechanism is fundamental to how verification, assigning truth, is managed in neural network theory. This is the predication. In logic a subject and predicate have no truth value, but it is their combination. In neural networks this is the relationship between any two things that can be recognized with one predicating another.
Recognizing a 'thing is referred to in language or other sensation' has no truth value in and of itself. It has to be understood in a predicative context to have truth, or survival value.
"Thing" can be an action like a lie, so don't think it has to be a physical object, it is a mental object. And any one mental object can predicate any other. Just as for any two nouns in any language, one can predicate the other.
This is precisely the mechanism of suspension of belief because neural computation is fundamentally discrimination or classification, not truth. A predication is a discrimination or classification of how one 'thing' modifies the understanding of another. In this sense, the question is not one of how does suspension of belief work, but how does actual belief, consideration that some imagining is true, work. Suspension of belief is fundamental to all neural computation because, precisely, it is evolved to take two different sensory/action patterns and regard them as the same. You recognize a "cup" despite never having it seen before. You recognize a "lie" despite never having heard, read, or seen it before. And, most importantly, you can assign truth by the framing context, the predication context, around that recognition, which is simply further classification/discrimination of the "cup" or "lie".
In seeing a "cup" or hearing a "lie" the immediately true predicating context is that you saw it, or you heard it. Other belief about that particular "cup" is then built upon with further predication framing. Recognizing other truths about it, or things that can be true about it.
In effect, truth is very hard for any brain to compute correctly. Teachable moments for real truth usually have to come from personal experience where you "learn the hard way." This is a selection among things you can understand by the fundamental neural computation of discrimination/classification which underlies all recognition whether of physical objects, physical actions, or ideas.
Another way to put this is that the brain is not very good at marking and remembering truth correctly because that is at odds with being able to classify things correctly. You see this in people not being able to immediately recall the color of a tie despite the fact that if the person had worn something that did not look like a tie but something odd, he would immediately trigger and remember that. Fundamentally, the brain always suspends belief, whether of objects or actions and can support elaborate verifications of exceptionally complicated predication structures never before encountered exactly that way before.
Belief happens when it has to, and not often. Most often it happens because of the hard wiring of brain computation to sense and action. Then it most easily happens in who do you trust. Animals readily recognize members of their own species which they may trust to be able to actually confirm truth they suspect (i.e., belief). Humans have the advantage of natural language which is completely unique in articulating computational-predication-alignment communications between brains. It is this uniquely evolved language capability communicating fundamentally non-linguistic brain computations that puts the possibility of enjoying fiction, or confirming actual truth, as a natural by-product of the fundamental brain computations.
It is truth that is not fundamental to brain computation. Truth is learned on mechanisms that fundamentally recognize things all of which will never, ever, repeat in brain signals exactly the same way twice. 'Suspension of belief' is what is fundamental to brain computation. Confirming belief that is not based on personal physical experience against hard reality is difficult for anybody unless they simply trust another person has figured it out already. That is what we observe when we see how easily people fall for fallacies as if true.
In computational psychology, inconsistency is always present. You never see exactly the same thing, the same electrical signals, when you see the exact same cup twice. You never see exactly the same thing when you hear or see the same kind of lie, twice. However, we are very good at using predications to scope truth and thereby resolve inconsistencies that may have been flagged in our recognition as direct recognitions of inconsistency.
This is why we directly recognize lies spoken. Just as directly as we recognize truth directly spoken. Whenever we recognize either. Bad fiction is not believable because the fiction writer has not taken this fundamental of understanding into account in his natural language communications of the tens of thousands if not millions of predications communicated, brain to brain, in a single book.
This article's historical analysis of suspension of belief is invaluable in helping to uncover the scientifically supported computational truth behind the intriguing fact that people love fiction. We now know the human brain is nothing like the computer as it is popularly understood today. Once all brain computation is understood in terms of what it actually computes, suspension of belief is fundamental and everywhere in thought. Human language is a window on brain computation that demonstrates belief is constantly suspended owing fundamentally and universally throughout the brain to the discrimination/classification functions which form the basis for predication and thought.
