# Show Reference: "Not Noisy, Just Wrong: The Role of Suboptimal Inference in Behavioral Variability"

Not noisy, just wrong: the role of suboptimal inference in behavioral variability. Neuron, Vol. 74, No. 1. (12 April 2012), pp. 30-39, doi:10.1016/j.neuron.2012.03.016 by Jeffrey M. Beck, Wei J. Ma, Xaq Pitkow, Peter E. Latham, Alexandre Pouget
@article{beck-et-al-2012,
abstract = {
Behavior varies from trial to trial even when the stimulus is maintained as constant as possible. In many models, this variability is attributed to noise in the brain. Here, we propose that there is another major source of variability: suboptimal inference. Importantly, we argue that in most tasks of interest, and particularly complex ones, suboptimal inference is likely to be the dominant component of behavioral variability. This perspective explains a variety of intriguing observations, including why variability appears to be larger on the sensory than on the motor side, and why our sensors are sometimes surprisingly unreliable.
},
author = {Beck, Jeffrey M. and Ma, Wei J. and Pitkow, Xaq and Latham, Peter E. and Pouget, Alexandre},
day = {12},
doi = {10.1016/j.neuron.2012.03.016},
issn = {1097-4199},
journal = {Neuron},
keywords = {ann, bayes, model, optimality, population-coding},
month = apr,
number = {1},
pages = {30--39},
pmid = {22500627},
posted-at = {2013-08-19 04:19:33},
priority = {2},
publisher = {Cell Press},
title = {Not noisy, just wrong: the role of suboptimal inference in behavioral variability.},
url = {http://dx.doi.org/10.1016/j.neuron.2012.03.016},
volume = {74},
year = {2012}
}


Beck et al. reinterpret Osborne et al.'s experiments stating it is more likely that sensory estimation in this task is suboptimal (thereby amplifying variability due to external and internal noise) than that the internal noise in perceptual and motor systems is vastly different.

Beck et al. acknowledge that the task in Osborne et al.'s experiments was very artificial and the brain circuits involved in smooth pursuit are probably optimized for more natural tasks.

Natural cognition is not always optimal.

Beck et al. argue that sub-optimal computations in biological and artificial neural networks can amplify behavioral and perceptual variability caused by internal and external noise.

Beck et al. argue that sub-optimal computations are a greater cause of behavioral and perceptual variability than internal noise.

Optimal operations are often not feasible for complex tasks for two reasons:

• the generative models necessary to do optimal estimation are too complex and require a lot of knowledge to create
• applying these models is much too computationally intensive

Optimal solutions to many computational tasks in perception and action have high computational complexity (in the complexity theory sense).

We cannot always expect optimal behavior in tasks which have become relevant only recently in human development, like eg. in complex reasoning tasks, or in tasks with highly artificial stimuli.

Tasks with high internal complexity can make it necessary to approximate optimal computations.

Such approximative computations can lead to highly suboptimal behavior even without internal or external noise.