Predictive coding is the neuroscience paradigm for researching the neural mechanisms of prediction and prediction error minimisation. The more than three decades of research in this space have shown that our brain is continuously engaged in the process of predicting its current and future states. Our brain does so by engaging broad hierarchical neural networks that operate across cortical and sub-cortical levels.

Research shows that within these large-scale predictive networks, our predictions are generated top-down from the higher-order cortical areas of the brain in response to sensory evidence. Research also indicates that the error minimisation processes occur as part of the same large-scale predictive networks, sending error signals back up the neural hierarchy for further processing.

Further Reading:

Disentangling predictive processing in the brain: a meta-analytic study in favour of a predictive network

Is there a prediction network? Meta-analytic evidence for a cortical-subcortical network likely subserving prediction.

The free energy principle is the computational neurobiological foundation underpinning predictive processing. Explained non-computationally, the free energy principle posits that we have a biological imperative to minimise free energy now and in the future (termed variational and expected free energy, respectively). This imperative to minimise free energy is achieved by reducing errors between our predictions and observations of incoming sensory data.

When we experience alignment between predictions and observations, functioning in the world is comparatively less demanding. In contrast, when we experience misalignments, we must mobilise free energy - in the form of resources such as neurochemical, emotional, attentional, and behavioural mechanisms - to help us move towards states of alignment. Aligned states are biologically preferred because they increase our chance of survival.

Further reading:

The free-energy principle: a unified brain theory?

Over short time scales, our predictive processes reflect adaptive behaviour. Over longer time scales, our capacity to better understand the causes of sensory information and develop more optimal models of our body in the world reflects learning.

Although learning often follows this temporal experiential process, we can also engage in bouts of learning. In such situations, we choose to expend energy (in the here and now) to explore and take in new information, which enables us to expand and refine our model of the world and minimise future prediction errors.

further reading:

Active inference and learning

Over the past decade, the predictive processing perspective has dramatically reconceptualised how researchers have come to think about emotions. In contrast to some traditional and subjective accounts of emotions being a reactive response to our experiences, emotions are now thought to be constructed from our predictions.

According to the theory of constructed emotion, our emotions are constructed (and categorised) based on our predictions (or best guess) of the causes of internal (interoceptive) and external (exteroceptive) sensory information, past experiences of similar sensory experiences, and beliefs about future states. Emotions, therefore, occur as part of the same predictive processes that underpin our thoughts and actions.

For example, our experience of the emotion fear stems from our best guess that the causes of sensory information likely suggest that there is a threat present (belief-based prediction) and that we should run and hide (action-based prediction).

Further reading:

Concepts Dissolve Artificial Boundaries in the Study of Emotion and Cognition, Uniting Body, Brain, and Mind

Research has started to look at play through the lens of predictive processing. This research suggests that play is an activity people can engage in to induce positive (feel-good) emotions whilst creating and resolving errors and establishing a better understanding of how the world works.

More specifically, the research suggests that play provides us with an opportunity to create and modify social, physical, and task-orientated environments in order to learn and understand “real-world” causal relationships - and, in doing so, establish more effective (or better predictive) models of the world for the future.

Further reading:

Play in predictive minds: A cognitive theory of play

Coming to grips with our world is not purely a process of minimising error in our current context. Instead, it is sometimes necessary to expend resources now in order to minimise potential prediction errors in future. This is what happens when we engage in long-term goal pursuit or learning and experiential activities.

In a similar vein, research suggests that creativity may emerge from the interplay between a predictive mind that is capable of imagining many different futures or scenarios and the environment. In such situations, we can form “exploration bubbles”, which allow us the space and time to test novel or out-of-the-box ideas and assess the evidence for or against them. The ultimate goal of engaging creativity is to expand our model of the world and take advantage of future opportunities.

Further reading:

Cultivating creativity: predictive brains and the enlightened room problem

Although we have the ability to act autonomously, we often find ourselves in situations where we are operating cohesively and collaboratively with others. Research suggests that the very same mechanisms that enable autonomous action support collective behaviour.

Specifically, over time, we become accustomed to the social and environmental forces in our context and seek to minimise prediction error in operating within the collective. Stated differently, it is often easier for us (in many contexts) to align our models of the world, predictions, and actions with others than to act individualistically.

Further reading:

Collective behavior from surprise minimization