top of page

An interactive model of divergent perception

Don't believe what you'll see? Click here to try ganzflicker

(pause video and read instructions first)

Click on the arrows inside the purple and orange dots to see the perceptual consequences of increasing and decreasing confidence in sensory input (S: purple dot) and mental imagery ability (M: orange dot). The size of the blue dot represents how strong your beliefs are (H: e.g., "What I'm seeing is meaningful") and the size of the green dot represents how strong your sensory expectations are (L: e.g., "the visual world has a predictable structure"). Confidence in sensory input affects the strength of beliefs and sensory expectations differently, which can lead to qualitatively different types of divergent experiences. Mental imagery ability affects the sensory richness of divergent experiences.

Like math? We use a general equation: g(μ) = β0 + F(X) to model complex (μ1), vivid simple (μ2), or non-vivid simple (μ3) pseudo-hallucinations: g(μ1i) = β0 + f1(SZi) + f2(1-Seni) + f3(Mi) g(μ2i) = β0 + f1(1-SZi) + f2(Seni) + f3(Mi) g(μ3i) = β0 + f1(1-SZi) + f2(Seni) + f3(1-Mi) where: β0 is the intercept term f is a specific function (used in place of F, denoting a group of functions) SZi is a measure of schizotypy (psychosis-proneness) Seni is a measure of sensory sensitivity (sensory confidence) Mi is a measure of mental imagery vividness

The divergent predictive perception model uses a predictive processing framework to demonstrate how confidence in sensory input influences the strength of high-level and low-level priors. Overly strong High-level priors (H: blue dot; beliefs such as "What I'm seeing is meaningful") result from low confidence in Sensory input (S: purple dot) - you are unlikely to update priors if you do not trust reality, which can lead to symptoms of psychosis such as hallucinations or delusions. The strength of Low-level priors (L: green dot; e.g., expectations about stimulus regularities) is inversely related to the strength of high-level priors (e.g., lack of trust in reality also breaks down expectations about how stimuli might appear in the real world). Overly strong low-level priors may make expectations narrow or inflexible, which could explain synaesthetic associations, which are automatic illusory experiences triggered consistently with other, veridical, experiences - such as seeing different letters of the alphabet in different colors. Synaesthesia is a form of divergent perception that is often enjoyable, not a diagnosable condition, and does not negatively impact quality of life - much unlike psychosis.


In our model, we explain how different factors contribute to different forms of divergent perception. We propose that the sensory richness of divergent experiences is related to mental imagery ability (M: orange dot). Combined with individual differences in sensory confidence, we can demonstrate different proneness to complex divergent perceptual experiences (e.g., hallucinations). The above model shows likelihoods of different divergent experiences (darker gray = higher likelihood), along with some sample percepts that might be experienced while observing ganzflicker, while under the influence of psychedelics, or in another setting that induces altered states of consciousness: the dog represents a complex hallucination, the rainbow pattern represents a vivid simple hallucination, and the dull pattern represents a non-vivid simple hallucination.


It is important to note that this model does not predict risk factors of developing psychiatric conditions. There is a wide normal range of individual differences in divergent experiences: healthy individuals can experience both hallucinations and reality discrimination issues from time to time.

This model is now published in Neuroscience of Consciousness (open access)!

bottom of page