Three experiments are reported that involve responding 10 the meaning or position of a word (ABOVE or BELOW) presented above or below a fixation point. Position and word meaning conflicted (ABOVElbelow or BELOW/above) or were compatible (ABOVE/above or BELOWlbelow), and the relative frequency of conflicting trials was varied. Experiment 1 required responses 10 the word and its position. Compatibility and frequency bad no effect in the spatial task, but interacted strongly in the word task: Compatible stimuli were processed faster when conflicting trials were rare (20% conflicting), but conflicting stimuli were processed faster when they were frequent (80% conflicting). Experiments 2 and 3 used the word task only and extended these findings to intermediate (20%, 40%, 60%, and 80% conflicting) and more extreme (10%, 20%, 80%, and 90% conflicting) frequencies, respectively. The advantage for conflicting stimuli when they were frequent was taken as evidence for a strategy involving dividing attention between reported and unreported dimensions. This paper reports an investigation of strategies for perforrning a Stroop-like task. It is based on the principle that performance in new task environments depends on a flexible yet predictable allocation of existing cognitive resources: The organization of existing resources can be changed quickly to do what the new task requires, and the strategy chosen will best exploit the regularities of the task environment to optimize performance (cf.

We define a process as autonomous (a) if it can begin without intention, and (b) if it can run on to completion without intention. We develop empirical criteria for determining whether a process can begin without intention, for determining whether it begins in the same way without intention as it does with intention, and for determining whether it can run on to completion without intention once it begins. We apply these criteria to assess the autonomy of the processes underlying simple mental arithmetic-the addition and multiplication of single digits-and find evidence that simple arithmetic may be only partially autonomous: It can begin without intention, but does not begin in the same way without intention as with intention and does not run on to completion without intention. This conclusion suggests there may be a continuum of autonomy, ranging from completely autonomous to completely nonautonomous.

In three experiments, human observers made timed decisions about alphanumeric characters, displayed singly in different orientations and versions (normal vs. backward). Latency to identify the characters was longer for backward than for normal versions, regardless of angular orientation and even under conditions in which latency was independent of angular orientation. Subjects also took longer to respond to a target orientation (whatever the character) than to respond to a target character (whatever the orientation). The results suggest that the observer first induces a description of a character that is largely independent of orientation but not of version, although the representation of version is too weak at this stage to permit an overt decision about it. Next, the angular orientation of the character is determined. Finally, the observer might "mentally rotate" the representation to the standard upright, for matching against an internally generated template.

In three experiments, subjects reported the identity of a word (ABOVE or BELOW) that appeared above or below a fixation point. On some trials, a cue presented 100, 200, 400, 600, 800, or 1,000 msec before the word indicated the relation between position and identity (i.e., whether the dimensions were compatible, e.g., ABOVE/above and BELOW/below, or conflicted, e.g., ABOVE/below and BELOW/ above). On the other trials, the cue was withheld (Experiment 2) or it bore no information about the relation between dimensions (Experiments 1 and 3). In each experiment, the cue reduced reaction time below the level of no-cue or neutral-cue controls, indicating strategic use of the relation between dimensions. Experiments 1 and 2 manipulated the number of potential cues that could occur in a block. A stronger cuing effect was found when one cue could occur (Experiment 2) than when two cues could occur (Experiment 1). Experiment 3 manipulated practice; it revealed that with practice the cuing effect reached asymptote at shorter delays. The asymptote itself did not change. Experiment 4 showed that cue-delay effects were independent of warning interval (warning interval and cue delay were confounded in Experiments 1, 2, and 3). The experiments demonstrate construction and utilization of strategies; they show that construction is sensitive to constraints imposed by the subject's goals and abilities and by the structure of the task environment.This article reports an investigation of goals in some task environment. The strategy strategies for performing a speeded discrim-chosen is probably a compromise between ination task. It is based on the premise that the constraints imposed by the informational nearly all cognitive tasks are performed as structures in the task environment, by the strategies, and that cognitive psychology can resources, capacities, abilities, or processes advance only by coming to understand what in the subject's cognitive repertoire, and by strategies are and how they are constructed the structure of the goals to be obtained by and utilized. In general, a strategy may be performing. This suggests that strategies defined as an optional organization of cog-may be investigated experimentally by manitive processes designed to achieve some nipulating the different constraints. This approach was adopted in the experiments re-" ported here.

Most theories of arithmetic assume that verification tasks are performed by producing an answer and comparing it with the presented answer. Verification is production plus comparison. We tested this hypothesis by imposing delays between arithmetic arguments and answers, in theory imposing delays between production and comparison. Long delays should absorb effects on production, and reaction time, from the onset of the answer, should reflect only comparison. Six experiments were conducted, three with addition and three with multiplication. Experiments 1 and 2 used experimenter-imposed delays; Experiments 3 and 4 used subject-imposed delays. In Experiments 5 and 6, subjects uttered the sum or product before exposing the answer. In Experiments 1-4, argument magnitude affected reaction time, even at the longest delay; in Experiments 5 and 6, argument magnitude effects were reduced. These results are contrary to the hypothesis that verification is production plus comparison and consistent with the idea that verification involves comparing the equation as a whole against memory.The psychology of simple arithmetic is based primarily on two main tasks, production, and verification. In production tasks, subjects are presented with a pair of digits and are asked to produce, usually by uttering aloud, their sum, product, or difference. For example, a subject presented with 3x5 = would be expected to utter "fifteen." By contrast, in verification tasks, subjects are presented with an equation containing a pair of digits and their putative sum, product, or difference and are asked to indicate whether the equation is true or false. For example, a subject presented with 3x5 = 17 would be expected to respond "false." This article concerns the relation between these tasks.In both tasks, reaction times vary substantially as the magnitudes of the digits are varied, and the patterns of variation are the primary evidence for models of the underlying representations and processes. The major competitors are counting models (Groen & Parkman, 1972), table-search models (Ashcraft & Battaglia, 1978;Geary, Widaman, & Little, 1986), and associative network models (Ashcraft, 1982(Ashcraft, , 1987Campbell, 1987aCampbell, , 1987bSiegler, 1988). In these theories, the magnitude of the digits (arguments) determines the amount of computation or the difficulty of retrieval, so argument magnitude is the major independent variable. The magnitude of the answer or the difference in the magnitude of the true answer and the presented answer is often treated as a secondary independent variable whose effects reflect processes that operate after computation or retrieval. These subsequent processes are a nuisance made necessary by the requirement