A little man of some importance

Abstract

The homunculus made his first appearance in the field of neurology on 1 December 1937, when Wilder Penfield and Edwin Boldrey (Fig. 1) published in Brain a 55-page article entitled Somatic motor and sensory representation in the cerebral cortex of man as studied by electrical stimulation (Penfield and Boldrey, 1937). The article is filled with painstakingly crafted summaries of data sourced from the cortical stimulation of 126 patients, who were operated under local anaesthesia by Penfield between 1928 and 1936. Compared to previous publications in animals, the authors had the advantage of operating on awake patients and relying on their verbal report of elicited movements and tactile sensation. All patient recordings were collated to obtain the first comprehensive map of motor and somatosensory localization in the human brain. This map was visualized as a distorted human-like figure—the homunculus—whose form indicates the amount of cortical area dedicated to motor or somatosensory functions of each body part. Wilder Penfield (left) and Edwin Boldrey (centre), creators of the homunculus (right). The paper had a long-lasting impact on both neurosurgical practice and scientific research. Clinically, it helped to establish direct cortical stimulation as an important method for functional mapping in neurosurgery; scientifically, it reinforced cortical localization as a valid paradigm to understand motor cognition, somatosensory perception and higher cognitive functions more generally. The homunculus was certainly their most striking proposition, but overshadowed other important conclusions made by the authors. First, while acknowledging the lack of histological analysis as a limitation of their study, they were confident they demonstrated the impossibility of confining ‘functional representation within strict cytoarchitectural boundaries’. Second, contrary to what others recorded in animals, the central sulcus of the human brain was not a clear boundary between motor and somatosensory areas. Many stimulations eliciting motor responses were indeed located in the somatosensory cortex of the postcentral gyrus and an even greater proportion of tactile responses in the motor cortex of the precentral gyrus. Surprisingly, the picture that emerged from their work was a clear degree of functional overlap between stimulation fields rather than an orderly sequence of segregated areas as their homunculus suggests. For the 80th anniversary of this seminal paper, the authors’ original findings are re-examined with the intent to clarify the substrate that generated the homunculus, arguably the most reproduced yet unreplicated figure in neuroscience. In the Middle Ages, a homunculus was an artificial humanoid created through alembics and mysterious alchemy. In modern times, Penfield and Boldrey brewed their own version of the homunculus using more controlled, though no less controversial, ingredients. Their approach was systematic and laborious. The neurological homunculus was generated from 170 summary maps of the number and location of stimulation points for each body part, each patiently sketched by Boldrey from Penfield’s operation notes, photographs and drawings. The maps that displayed the positive responses for the toes, legs, trunk, arm, digits, hand, face, eyes, mouth and tongue were used to extract three separate measures for the motor (Fig. 2) and the somatosensory (Supplementary Fig. 1) stimulations: (i) the area for each body part as a 2D surface displayed on the lateral aspect of the central fronto-parietal cortex (Fig. 2B); (ii) the count of the total number of stimulations located in front and posterior to the central sulcus (Fig. 2C); and (iii) the vertical extent of each body area along the central sulcus (Fig. 2D). Representation of the motor stimulations for different body parts. (A) Colour-coding of different body parts. (B) Areas of the surface maps enclosing all motor stimulation points for each body part. (C) Count of the number of stimulations and (D) measurements of the vertical length of the surface maps for each body part. Original data are derived from Penfield and Boldrey (1937). The asterisk indicates the impossibility of generating any measurement of the area and length from the few stimulations recorded for the trunk. Similar maps for the somatosensory stimulations are provided in Supplementary Fig. 1. In their attempt to visually summarize such a large amount of data, the authors faced challenges that are common to modern neuroimaging approaches (Friston et al., 2004). First, the problem of quality control and data transformation, a step that in their case included elimination of artefactual stimulations (e.g. due to epileptic activity, shift in the proximity of the tumour, inability to replicate an evoked response, etc.) and spatial normalization to a common template. These steps were performed solely by visual inspection and manual transcription, processes that are vulnerable to error as evi