The interactive race model of saccadic countermanding assumes that response inhibition results from an interaction between a go unit, identified with gaze-shifting neurons, and a stop unit, identified with gaze-holding neurons, in which activation of the stop unit inhibits the growth of activation in the go unit to prevent it from reaching threshold. The interactive race model accounts for behavioral data and predicts physiological data in monkeys performing the stop-signal task. We propose an alternative model that assumes that response inhibition results from blocking the input to the go unit. We show that the blocked-input model accounts for behavioral data as accurately as the original interactive race model and predicts aspects of the physiological data more accurately. We extend the models to address the steady-state fixation period before the go stimulus is presented and find that the blocked-input model fits better than the interactive race model. We consider a model in which fixation activity is boosted when a stop signal occurs and find that it fits as well as the blocked input model but predicts very high steady-state fixation activity after the response is inhibited. We discuss the alternative linking propositions that connect computational models to neural mechanisms, the lessons to be learned from model mimicry, and generalization from countermanding saccades to countermanding other kinds of responses.

Theories of skilled performance propose that highly trained skills involve hierarchically structured control processes. The present study examined and demonstrated hierarchical control at several levels of processing in skilled typewriting. In the first two experiments, we scrambled the order of letters in words to prevent skilled typists from chunking letters, and compared typing words and scrambled words. Experiment 1 manipulated stimulus quality to reveal chunking in perception, and Experiment 2 manipulated concurrent memory load to reveal chunking in short-term memory (STM). Both experiments manipulated the number of letters in words and nonwords to reveal chunking in motor planning. In the next two experiments, we degraded typing skill by altering the usual haptic feedback by using a laser-projection keyboard, so that typists had to monitor keystrokes. Neither the number of motor chunks (Experiment 3) nor the number of STM items (Experiment 4) was influenced by the manipulation. The results indicate that the utilization of hierarchical control depends on whether the input allows chunking but not on whether the output is generated automatically. We consider the role of automaticity in hierarchical control of skilled performance.

Associations between corresponding stimulus-response locations are often characterized as overlearned, producing automatic activation. However, 84 practice trials with an incompatible mapping eliminate the benefit for spatial correspondence in a transfer Simon task, where stimulus location is irrelevant. The authors examined whether transfer occurs for combinations of physical-location, arrow-direction, and location-word modes in the practice and transfer sessions. With 84 practice trials, the Simon effect was reduced for locations and arrows, and there was complete transfer across these modes; location words showed little transfer within or between modes. These results suggest that the acquired short-term associations were based on visual-spatial stimulus codes distinct from semantic-spatial codes activated by the words. With 600 practice trials, words showed transfer to word and arrow but not location Simon tasks, suggesting that arrows share semantic-spatial codes with words. Reaction-time distribution functions for the Simon effect showed distinct shapes for each stimulus mode, with little impact of the practiced mapping on the shapes. Thus, the contribution of the short-term location associations seems to be separate from that of the long-term associations responsible for the Simon effect.

The present study examined the stimulus-response compatibility (SRC) effect in a simulated flight environment. Experiments 1 and 2 tested the effect with pure and mixed mappings in flight tasks by using attitude displays with inside-out and outside-in formats, whereas Experiments 3 and 4 used a simplified display and tasks. The SRC effect was obtained with mixed mappings when responses were turns of a flight yoke (Experiments 1-3). In contrast, the SRC effect was absent with mixed mappings when they were buttonpresses (Experiment 4). Analyses of sequential effects suggest that the reduction in Experiments 1-3 can be attributed to reduction in the frequency of trials for which the congruent mapping repeats, but the elimination in Experiment 4 cannot be. Implications of the findings are discussed in the context of aviation cockpit design.

Typing performance involves hierarchically structured control systems: At the higher level, an outer loop generates a word or a series of words to be typed; at the lower level, an inner loop activates the keystrokes comprising the word in parallel and executes them in the correct order. The present experiments examined contributions of the outer-and inner-loop processes to the control of speed and accuracy in typewriting. Experiments 1 and 2 involved discontinuous typing of single words, and Experiments 3 and 4 involved continuous typing of paragraphs. Across experiments, typists were able to trade speed for accuracy but were unable to type at rates faster than 100 ms/keystroke, implying limits to the flexibility of the underlying processes. The analyses of the component latencies and errors indicated that the majority of the trade-offs were due to inner-loop processing. The contribution of outer-loop processing to the trade-offs was small, but it resulted in large costs in error rate. Implications for strategic control of automatic processes are discussed.Keywords: motor control, hierarchical processing, speed-accuracy trade-off, skilled performance, action sequence A major issue in studies of cognitive skill concerns the ways in which skilled performance can be controlled strategically. Skill is acquired through training, and the extensive training required to attain high levels of skill makes component processes highly automatic (Anderson, 1982;Fitts, 1964;Logan, 1988;Schneider & Shiffrin, 1977); practice makes them specialized, stereotyped, and inflexible. Paradoxically, expert skills are often robust and flexible, so they can be utilized in various task contexts (MacKay, 1982). The present study addresses this paradox, asking whether automatic processes can be controlled strategically. To this end, we investigated a basic form of strategic control in cognitive performance-the trade-off between speed and accuracy. We focused on typewriting, which is one of the most prevalent skills among college students in modern society (Logan & Crump, 2011). Hierarchical Control in Skilled TypingTypewriting is an ideal subject for investigation of skilled performance. College students typically have a semester of formal training in middle school and 10 to 11 years of experience in which they type every day (Logan & Crump, 2011). Typing is a complex skill: It involves selecting specific keys that correspond to the letters in a to-be-typed word, moving appropriate fingers to precise key locations, and executing keystrokes in the correct order. Skilled typists implement these processes rapidly without watching the fingers that they control (i.e., touch typing).Typewriting is an expression of language, and language has hierarchically nested structures: Texts contain sentences, sentences contain words, and words contain letters. Skilled typewriting involves hierarchically organized cognitive processes that address these levels of representation (Fendrick, 1937;Lashley, 1951;Logan & Crump, 2011;Shaffer, 1975;Sternberg, Knoll, &...