Limits...
Active inference, sensory attenuation and illusions.

Brown H, Adams RA, Parees I, Edwards M, Friston K - Cogn Process (2013)

Bottom Line: Furthermore, it explains the force-matching illusion and reproduces empirical results almost exactly.This is important, given the negative correlation between sensory attenuation and delusional beliefs in normal subjects--and the reduction in the magnitude of the illusion in schizophrenia.It also provides a functional account of deficits in syndromes characterised by false inference and impaired movement--like schizophrenia and Parkinsonism--syndromes that implicate abnormal modulatory neurotransmission.

View Article: PubMed Central - PubMed

Affiliation: Institute of Neurology, The Wellcome Trust Centre for Neuroimaging, UCL, 12 Queen Square, London, WC1N 3BG, UK, harriet.brown.09@ucl.ac.uk.

ABSTRACT
Active inference provides a simple and neurobiologically plausible account of how action and perception are coupled in producing (Bayes) optimal behaviour. This can be seen most easily as minimising prediction error: we can either change our predictions to explain sensory input through perception. Alternatively, we can actively change sensory input to fulfil our predictions. In active inference, this action is mediated by classical reflex arcs that minimise proprioceptive prediction error created by descending proprioceptive predictions. However, this creates a conflict between action and perception; in that, self-generated movements require predictions to override the sensory evidence that one is not actually moving. However, ignoring sensory evidence means that externally generated sensations will not be perceived. Conversely, attending to (proprioceptive and somatosensory) sensations enables the detection of externally generated events but precludes generation of actions. This conflict can be resolved by attenuating the precision of sensory evidence during movement or, equivalently, attending away from the consequences of self-made acts. We propose that this Bayes optimal withdrawal of precise sensory evidence during movement is the cause of psychophysical sensory attenuation. Furthermore, it explains the force-matching illusion and reproduces empirical results almost exactly. Finally, if attenuation is removed, the force-matching illusion disappears and false (delusional) inferences about agency emerge. This is important, given the negative correlation between sensory attenuation and delusional beliefs in normal subjects--and the reduction in the magnitude of the illusion in schizophrenia. Active inference therefore links the neuromodulatory optimisation of precision to sensory attenuation and illusory phenomena during the attribution of agency in normal subjects. It also provides a functional account of deficits in syndromes characterised by false inference and impaired movement--like schizophrenia and Parkinsonism--syndromes that implicate abnormal modulatory neurotransmission.

Show MeSH

Related in: MedlinePlus

Generative model: This figure shows the generative process and model used in these simulations. The generative process (left) models real-world states and causes, while the generative model (right) is used by the subject to make inferences about causes of its sensations. In the real world, the hidden state xi models self-generated forces that are sensed by both somatosensory ss and proprioceptive sp input channels. External forces are modelled with the hidden cause νe and are sensed only by the somatosensory input channel. Action causes the self-generated force to increase and is modified by a sigmoid squashing function σ (a hyperbolic tangent function). The hidden state decays slowly over four time bins. In the generative model, causes of sensory data are divided into internal causes νi and external causes νe. The hidden cause excites dynamics in hidden states xi and xe which decay slowly over time as above. Internal force is perceived by both proprioceptive and somatosensory receptors, while external force is perceived only by somatosensory receptors. Crucially, the precision of the sensory prediction error π is influenced by the level of internal force, again modulated by a squashing function, and controlled by a parameter γ which governs the level of attenuation of precision. The pink circles highlight this state-dependent precision, which effectively controls the influence of sensory prediction errors during active inference
© Copyright Policy - OpenAccess
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3824582&req=5

Fig1: Generative model: This figure shows the generative process and model used in these simulations. The generative process (left) models real-world states and causes, while the generative model (right) is used by the subject to make inferences about causes of its sensations. In the real world, the hidden state xi models self-generated forces that are sensed by both somatosensory ss and proprioceptive sp input channels. External forces are modelled with the hidden cause νe and are sensed only by the somatosensory input channel. Action causes the self-generated force to increase and is modified by a sigmoid squashing function σ (a hyperbolic tangent function). The hidden state decays slowly over four time bins. In the generative model, causes of sensory data are divided into internal causes νi and external causes νe. The hidden cause excites dynamics in hidden states xi and xe which decay slowly over time as above. Internal force is perceived by both proprioceptive and somatosensory receptors, while external force is perceived only by somatosensory receptors. Crucially, the precision of the sensory prediction error π is influenced by the level of internal force, again modulated by a squashing function, and controlled by a parameter γ which governs the level of attenuation of precision. The pink circles highlight this state-dependent precision, which effectively controls the influence of sensory prediction errors during active inference

Mentions: Figure 1 describes the generative process and model in terms of equations (that have the same hierarchical form as Eq. 2) and a schematic showing how the hidden states and causes are interpreted. This model is as simple as we could make it, while retaining the key ingredients that are required to demonstrate inference about or attribution of agency. The equations on the left describe the real world (whose states and causes are in boldface), while the equations on the right constitute the subject’s generative model. In the real world, there is one hidden state modelling self-generated force or pressure that is registered by both proprioceptive and somatosensory input. This hidden force increases with action and decays with a time constant of four time bins (where each time bin corresponds to about 100 ms). Externally generated forces are modelled with and add to the internally generated forces to provide somatosensory input.Fig. 1


Active inference, sensory attenuation and illusions.

Brown H, Adams RA, Parees I, Edwards M, Friston K - Cogn Process (2013)

Generative model: This figure shows the generative process and model used in these simulations. The generative process (left) models real-world states and causes, while the generative model (right) is used by the subject to make inferences about causes of its sensations. In the real world, the hidden state xi models self-generated forces that are sensed by both somatosensory ss and proprioceptive sp input channels. External forces are modelled with the hidden cause νe and are sensed only by the somatosensory input channel. Action causes the self-generated force to increase and is modified by a sigmoid squashing function σ (a hyperbolic tangent function). The hidden state decays slowly over four time bins. In the generative model, causes of sensory data are divided into internal causes νi and external causes νe. The hidden cause excites dynamics in hidden states xi and xe which decay slowly over time as above. Internal force is perceived by both proprioceptive and somatosensory receptors, while external force is perceived only by somatosensory receptors. Crucially, the precision of the sensory prediction error π is influenced by the level of internal force, again modulated by a squashing function, and controlled by a parameter γ which governs the level of attenuation of precision. The pink circles highlight this state-dependent precision, which effectively controls the influence of sensory prediction errors during active inference
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3824582&req=5

Fig1: Generative model: This figure shows the generative process and model used in these simulations. The generative process (left) models real-world states and causes, while the generative model (right) is used by the subject to make inferences about causes of its sensations. In the real world, the hidden state xi models self-generated forces that are sensed by both somatosensory ss and proprioceptive sp input channels. External forces are modelled with the hidden cause νe and are sensed only by the somatosensory input channel. Action causes the self-generated force to increase and is modified by a sigmoid squashing function σ (a hyperbolic tangent function). The hidden state decays slowly over four time bins. In the generative model, causes of sensory data are divided into internal causes νi and external causes νe. The hidden cause excites dynamics in hidden states xi and xe which decay slowly over time as above. Internal force is perceived by both proprioceptive and somatosensory receptors, while external force is perceived only by somatosensory receptors. Crucially, the precision of the sensory prediction error π is influenced by the level of internal force, again modulated by a squashing function, and controlled by a parameter γ which governs the level of attenuation of precision. The pink circles highlight this state-dependent precision, which effectively controls the influence of sensory prediction errors during active inference
Mentions: Figure 1 describes the generative process and model in terms of equations (that have the same hierarchical form as Eq. 2) and a schematic showing how the hidden states and causes are interpreted. This model is as simple as we could make it, while retaining the key ingredients that are required to demonstrate inference about or attribution of agency. The equations on the left describe the real world (whose states and causes are in boldface), while the equations on the right constitute the subject’s generative model. In the real world, there is one hidden state modelling self-generated force or pressure that is registered by both proprioceptive and somatosensory input. This hidden force increases with action and decays with a time constant of four time bins (where each time bin corresponds to about 100 ms). Externally generated forces are modelled with and add to the internally generated forces to provide somatosensory input.Fig. 1

Bottom Line: Furthermore, it explains the force-matching illusion and reproduces empirical results almost exactly.This is important, given the negative correlation between sensory attenuation and delusional beliefs in normal subjects--and the reduction in the magnitude of the illusion in schizophrenia.It also provides a functional account of deficits in syndromes characterised by false inference and impaired movement--like schizophrenia and Parkinsonism--syndromes that implicate abnormal modulatory neurotransmission.

View Article: PubMed Central - PubMed

Affiliation: Institute of Neurology, The Wellcome Trust Centre for Neuroimaging, UCL, 12 Queen Square, London, WC1N 3BG, UK, harriet.brown.09@ucl.ac.uk.

ABSTRACT
Active inference provides a simple and neurobiologically plausible account of how action and perception are coupled in producing (Bayes) optimal behaviour. This can be seen most easily as minimising prediction error: we can either change our predictions to explain sensory input through perception. Alternatively, we can actively change sensory input to fulfil our predictions. In active inference, this action is mediated by classical reflex arcs that minimise proprioceptive prediction error created by descending proprioceptive predictions. However, this creates a conflict between action and perception; in that, self-generated movements require predictions to override the sensory evidence that one is not actually moving. However, ignoring sensory evidence means that externally generated sensations will not be perceived. Conversely, attending to (proprioceptive and somatosensory) sensations enables the detection of externally generated events but precludes generation of actions. This conflict can be resolved by attenuating the precision of sensory evidence during movement or, equivalently, attending away from the consequences of self-made acts. We propose that this Bayes optimal withdrawal of precise sensory evidence during movement is the cause of psychophysical sensory attenuation. Furthermore, it explains the force-matching illusion and reproduces empirical results almost exactly. Finally, if attenuation is removed, the force-matching illusion disappears and false (delusional) inferences about agency emerge. This is important, given the negative correlation between sensory attenuation and delusional beliefs in normal subjects--and the reduction in the magnitude of the illusion in schizophrenia. Active inference therefore links the neuromodulatory optimisation of precision to sensory attenuation and illusory phenomena during the attribution of agency in normal subjects. It also provides a functional account of deficits in syndromes characterised by false inference and impaired movement--like schizophrenia and Parkinsonism--syndromes that implicate abnormal modulatory neurotransmission.

Show MeSH
Related in: MedlinePlus