Unconvincing statistical and functional inferences: reply to Catmur

摘要: In a recent commentary published in Frontiers Human Neuroscience, Catmur (2014) raises several important questions for discussion about study we earlier this year (Michael et al., 2014). the following, however, point out that her criticism is based upon two inferences do not find convincing, and maintain conservative interpretation offered our original article more appropriate than Catmur's alternative interpretation. In study, used offline continuous theta-burst stimulation (cTBS) to investigate whether regions of premotor cortex (PMC) play causal role action understanding. Participants received cTBS over hand lip areas left PMC, separate sessions, before completing pantomime-recognition task which half trials contained pantomimed actions, mouth actions. The results revealed double dissociation: were less accurate recognizing actions after receiving area area. We argued finding constrains theories understanding by showing somatotopically organized PMC contribute causally understanding. However, as observes, further aim was shed light specific functional targeted neural populations, i.e., they encoding kinematic information observed or proximal distal goals (e.g., grasp cup drink) those therefore devised three tasks probe different components simplest required participants identify still frames from brief videos This thus probed perceptual component understanding, ability process features An intermediate them select objects complemented video action, probing their goal an action. most complex context-sensitive manner, showed no significant difference among these tasks. concludes adequate data would be one identifies contribution populations whatever common tasks—and information. As she puts it: “Recall all perception: since performance on impaired, suggest mirror neuron are involved, higher-level processes such matching its object selecting relevant given context, but instead lower-level perception” (Catmur, 2014, p. 2). Note, constitutes inference absence evidence absence. But interaction does warrant impact magnetic same; it simply means did different. may have been due various reasons, candidate being lack statistical power. A possibility any neurons contributing differently levels complexity distributed, affected equally at sites. view ignorance matter, believe conclusion originally (acknowledging specialize processing low-level information) one: “Given three-way TMS site, type, level (simple, inter-mediate, complex), permit hierarchical (i.e., specifically encode kinematics, goals, goals). One movements varying complexity” 970). It also worth noting draws second well: claim involved “action “higher-level context.” depends radically modularist premise possible some underpinning perception sometimes making judgments (proximal distal) see reason accept premise. result, while agree with assessment question possibly (even likely) relatively (encoding actions), can actions. Clearly there risk falling into mere terminological quibbling terms understanding,” right observe has hampered research It part order avoid invoked precise tripartite distinction [borrowed Hamilton Grafton (2007)]. Although ultimately resolve attempt operationalize useful starting research.