Subtopic Deep Dive

Internal Models in Sensorimotor Control
Research Guide

What is Internal Models in Sensorimotor Control?

Internal models in sensorimotor control are neural representations that predict sensory consequences of motor commands (forward models) and compute motor commands from desired sensory outcomes (inverse models).

The brain uses forward models to anticipate movement outcomes and inverse models to plan actions, enabling rapid adaptation to novel dynamics. Foundational work by Shadmehr and Mussa-Ivaldi (1994, 2637 citations) showed adaptive representation of dynamics in reaching tasks. Over 10 key papers from 1994-2012 explore visuomotor transformations, implicit learning, and active inference (Krakauer et al., 2000, 910 citations; Friston et al., 2010, 833 citations).

15
Curated Papers
3
Key Challenges

Why It Matters

Internal models explain predictive control in motor adaptation, informing neuroprosthetics design where forward models predict limb states from prosthetic signals (Flanagan and Wing, 1997). They underpin computational neuroscience simulations of cerebellar function in error-based learning (Shadmehr and Brashers-Krug, 1997). Free-energy formulations link models to active inference for robotics and brain-machine interfaces (Friston et al., 2010; Adams et al., 2012).

Key Research Challenges

Separating Forward from Inverse Models

Distinguishing contributions of forward and inverse models in adaptation remains difficult due to overlapping behavioral signatures. Shadmehr and Mussa-Ivaldi (1994) demonstrated adaptive dynamics learning but could not isolate model types cleanly. Krakauer et al. (2000) highlighted vectorial planning differences, complicating inverse model identification.

Implicit vs. Explicit Learning Processes

Implicit internal model updates override explicit strategies, as shown in visuomotor tasks (Mazzoni and Krakauer, 2006, 872 citations). This hierarchy challenges models of conscious motor control. Resolving neural substrates requires integrating behavioral and neuroimaging data (Kelly and Garavan, 2004).

Neural Implementation of Active Inference

Linking free-energy predictions to motor cortex hierarchies lacks direct evidence (Adams et al., 2012). Friston et al. (2010) proposed variational inference but empirical validation in humans is sparse. Timing and distributed systems add complexity (Rao et al., 1997).

Essential Papers

1.

Adaptive representation of dynamics during learning of a motor task

Reza Shadmehr, FA Mussa-Ivaldi · 1994 · Journal of Neuroscience · 2.6K citations

We investigated how the CNS learns to control movements in different dynamical conditions, and how this learned behavior is represented. In particular, we considered the task of making reaching mov...

2.

Learning of Visuomotor Transformations for Vectorial Planning of Reaching Trajectories

John W. Krakauer, Zachary M. Pine, M. F. Ghilardi et al. · 2000 · Journal of Neuroscience · 910 citations

The planning of visually guided reaches is accomplished by independent specification of extent and direction. We investigated whether this separation of extent and direction planning for well pract...

3.

An Implicit Plan Overrides an Explicit Strategy during Visuomotor Adaptation

Pietro Mazzoni, John W. Krakauer · 2006 · Journal of Neuroscience · 872 citations

The relationship between implicit and explicit processes during motor learning, and for visuomotor adaptation in particular, is poorly understood. We set up a conflict between implicit and explicit...

4.

Action and behavior: a free-energy formulation

Karl Friston, Jean Daunizeau, James M. Kilner et al. · 2010 · Biological Cybernetics · 833 citations

We have previously tried to explain perceptual inference and learning under a free-energy principle that pursues Helmholtz's agenda to understand the brain in terms of energy minimization. It is fa...

5.

Predictions not commands: active inference in the motor system

Rick A. Adams, Stewart Shipp, Karl Friston · 2012 · Brain Structure and Function · 780 citations

The descending projections from motor cortex share many features with top-down or backward connections in visual cortex; for example, corticospinal projections originate in infragranular layers, ar...

6.

Human Functional Neuroimaging of Brain Changes Associated with Practice

Clare Kelly, Hugh Garavan · 2004 · Cerebral Cortex · 741 citations

The discovery that experience-driven changes in the human brain can occur from a neural to a cortical level throughout the lifespan has stimulated a proliferation of research into how neural functi...

7.

Functional Stages in the Formation of Human Long-Term Motor Memory

Reza Shadmehr, Thomas Brashers-Krug · 1997 · Journal of Neuroscience · 720 citations

Previous research has demonstrated that the primate CNS has the ability to learn and store multiple and conflicting visuo-motor maps. Here we studied the ability of human subjects to learn to make ...

Reading Guide

Foundational Papers

Start with Shadmehr and Mussa-Ivaldi (1994) for adaptive dynamics core; Krakauer et al. (2000) for visuomotor planning; Mazzoni and Krakauer (2006) for implicit-explicit distinction.

Recent Advances

Friston et al. (2010) on free-energy action; Adams et al. (2012) on predictive motor inference.

Core Methods

Force-field reaching tasks, visuomotor rotations, grip force prediction, fMRI timing studies, free-energy minimization.

How PapersFlow Helps You Research Internal Models in Sensorimotor Control

Discover & Search

Research Agent uses citationGraph on Shadmehr and Mussa-Ivaldi (1994) to map 2637 citing works, revealing clusters in cerebellar adaptation; exaSearch queries 'forward internal models visuomotor rotation' to find Krakauer et al. (2000) and similar papers; findSimilarPapers expands from Friston et al. (2010) to active inference extensions.

Analyze & Verify

Analysis Agent runs readPaperContent on Mazzoni and Krakauer (2006) to extract implicit override data, then verifyResponse with CoVe against behavioral claims; runPythonAnalysis replots force field adaptation curves from Shadmehr (1994) using NumPy for statistical verification; GRADE grading scores evidence strength in adaptation studies.

Synthesize & Write

Synthesis Agent detects gaps in inverse model evidence across Flanagan (1997) and Krakauer (2000), flagging contradictions in explicit strategies; Writing Agent applies latexEditText to draft model equations, latexSyncCitations for 10+ papers, and latexCompile for publication-ready reviews; exportMermaid visualizes forward-inverse model hierarchies.

Use Cases

"Replot adaptation curves from Shadmehr 1994 force field learning with Python"

Research Agent → searchPapers 'Shadmehr Mussa-Ivaldi 1994' → Analysis Agent → readPaperContent → runPythonAnalysis (NumPy/matplotlib replots dynamics curves, computes error metrics) → researcher gets annotated plots and stats.

"Write LaTeX review on implicit vs explicit models citing Mazzoni Krakauer"

Research Agent → citationGraph 'Mazzoni Krakauer 2006' → Synthesis → gap detection → Writing Agent → latexEditText (drafts section) → latexSyncCitations (adds 872+ refs) → latexCompile → researcher gets PDF with equations.

"Find code for simulating internal models in reaching tasks"

Research Agent → searchPapers 'visuomotor internal models simulation' → Code Discovery → paperExtractUrls → paperFindGithubRepo → githubRepoInspect → researcher gets runnable forward model code from adaptation repos.

Automated Workflows

Deep Research workflow scans 50+ citing papers to Shadmehr (1994) via citationGraph, producing structured review with GRADE scores on model evidence. DeepScan applies 7-step CoVe to verify active inference claims in Friston (2010) and Adams (2012). Theorizer generates hypotheses linking cerebellar timing (Rao 1997) to forward model updates.

Frequently Asked Questions

What defines forward and inverse internal models?

Forward models predict sensory outcomes from motor commands; inverse models compute commands for desired outcomes (Shadmehr and Mussa-Ivaldi, 1994).

What methods test internal models?

Force field adaptation and visuomotor rotations probe model updates; grip force adjustments reveal predictions (Flanagan and Wing, 1997; Krakauer et al., 2000).

What are key papers?

Shadmehr and Mussa-Ivaldi (1994, 2637 citations) on dynamics; Mazzoni and Krakauer (2006, 872 citations) on implicit plans; Friston et al. (2010, 833 citations) on free-energy.

What open problems exist?

Isolating model types neurally; integrating active inference with timing networks; validating in neuroprosthetics (Adams et al., 2012; Rao et al., 1997).

Research Motor Control and Adaptation with AI

PapersFlow provides specialized AI tools for Neuroscience researchers. Here are the most relevant for this topic:

See how researchers in Life Sciences use PapersFlow

Field-specific workflows, example queries, and use cases.

Life Sciences Guide

Start Researching Internal Models in Sensorimotor Control with AI

Search 474M+ papers, run AI-powered literature reviews, and write with integrated citations — all in one workspace.

See how PapersFlow works for Neuroscience researchers