Working-memory capacity and the control of attention: The contributions of goal neglect, response competition, and task set to Stroop interference.

In tasks with no or few congruent trials, or in high-congruency tasks that followed low-congruency tasks, WM predicted response-time interference.

“In contrast, in tasks with no or few congruent trials, or in high-congruency tasks that followed low-congruency tasks, WM predicted response-time interference.”

Working-memory capacity differences predicted performance on the Stroop task across five experiments.

“Individual differences in working-memory (WM) capacity predicted performance on the Stroop task in 5 experiments”

WM was related to response latency, not accuracy, in contexts that reinforced the task goal and minimized the difficulty of actively maintaining it.

“WM was related to latency, not accuracy, in contexts that reinforced the task goal and so minimized the difficulty of actively maintaining it.”

When a Stroop task included many congruent trials, low WM individuals made more errors than high WM individuals on the rare incongruent trials that required maintaining the ignore-word goal.

“When the Stroop task encouraged goal neglect by including large numbers of congruent trials (RED presented in red), low WM individuals committed more errors than did high WM individuals on the rare incongruent trials (BLUE in red) that required maintaining access to the -ignore-the-word‖ goal for accurate responding.”

• AssumptionRests on an assumption: Participants complied with the instruction to ignore the word on every trial and did not look away or squint during the Stroop task.

Low spans did not show larger interference effects than high spans overall in Experiment 1.

“low spans did not show larger interference effects than did high spans, either overall, F(1, 83) = 2.22, MSE = 2,669.56,p = .14”

In Experiment 2, high spans had shorter naming latencies overall than low spans.

“although high spans had shorter naming latencies than did low spans overall, F(1, 86) = 11.40,MSE = 45,504.65”

In Experiments 1 and 2, 11% and 13% of high spans, respectively, made at least 20% errors.

“In Experiments 1 and 2, 11% and 13% of high spans, respectively, made ≥ 20% errors.”

In Experiment 1, contrary to typical Stroop findings, span differences appeared in accuracy rather than response time.

“But contrary to prototypical Stroop findings, these span differences arose in accuracy, and not in response time.3”

Different manifestations of Stroop interference are dissociable by their relation to WM capacity.

“The data indicate that different manifestations of Stroop interference are dissociable by their relation to WM capacity.”

In Experiment 2, response-time facilitation was significant for each span group.

“(again, the facilitation effect was significant for each span group, with Fs > 40).”

In Experiment 2's 75% congruent condition, error interference was significant for each span group.

“(the error interference effect was significant for each span group, with Fs > 26).”

In Experiment 2, despite accuracy feedback after every trial, low spans still made more errors on incongruent trials than high spans in the 75% congruent task.

“Despite this salient feedback provided after every trial, low spans still committed more errors on incongruent trials than did high spans in the 75% congruent task.”

In Experiment 1, low spans made more errors overall than high spans.

“Low spans committed more errors overall than did high spans, F(1, 83) = 5.73, MSE = 0.004”

In Experiment 1's 0% congruent condition, the error-based Stroop effect was significant overall.

“In the 0% congruent condition, the Stroop effect was significant overall, F(1, 41) = 14.19,MSE = 0.001”

In Experiment 1's 75% congruent condition, low-span participants showed larger response-time facilitation than high-span participants.

“Moreover, low spans appeared to show larger facilitation effects than did high spans in the 75% congruent condition.”

• Also stated in: § Response Times · plain

  “this facilitation was greater for low spans than for high spans, F(1, 42) = 3.63, MSE = 889.97,p = .06”

High- and low-span participants showed similar naming latencies overall in Experiment 1.

“High-and low-span participants showed similar naming latencies overall, F(1, 83) = 1.16,MSE = 24,695.07,p = .28”

In Experiment 1's 75% congruent condition, there was a significant overall error-based Stroop interference effect.

“the 75% congruent condition produced an overall Stroop interference effect, F(1, 42) = 61.63,MSE = 0.005”

The locus of WM span differences in Stroop interference shifts between response latencies and error rates depending on the set induced by the current and previous congruency context.

“That is, the locus of WM span differences in Stroop interference is systematically shifted between response latencies and error rates depending on the set induced by the current, and the previous, -congruency‖ context.”

In Experiment 1, low spans showed a larger error-based Stroop interference effect than high spans.

“critically, low spans showed a larger Stroop interference effect than did high spans, F(1, 83) = 7.56, MSE = 0.003.”

Presenting the 0% congruent task after the 75% task did not appreciably change the results.

“Presenting the 0% task after the 75% task did not appreciably change the results.”

In Experiment 1, error rates were sensitive to span differences in interference only in the 75% congruent condition.

“Inspection of the means suggests that, unlike response times, error rates were highly sensitive to span differences in interference, but this was true for the 75% congruent condition only.”

In Experiment 1's 0% congruent context, no WM-related differences appeared in either response-time or error interference.

“Thus, in the 0% congruent context, no WM-related differences were seen in either response time or error”

In Experiment 2, low spans showed larger error interference than high spans only in the 75% congruent condition.

“Inspection of the means in Table 2 suggests that low spans showed a larger error interference effect than did high spans in the 75% congruent condition only, as they had in Experiment 1.”

In Experiment 1, the span difference in error interference was larger in the 75% congruent condition than in the 0% condition.

“span interacted with trial type and congruency proportion, F(1, 83) = 3.94, MSE = 0.003, p = .05,indicating that the span difference in interference was larger in the 75% congruent condition than in the 0% condition.”

In Experiment 1's 0% congruent condition, high and low spans did not differ in error interference.

“but high and low spans did not differ in interference, F(1, 41) = 1.08,MSE = 0.001,p = .31.”

In Experiment 1, 28% of low spans and no high spans had error rates over 30%.

“error rates over 30% were demonstrated by 28% of low spans and no high spans.”

In Experiment 1, error interference was larger in the 75% congruent condition than in the 0% congruent condition.

“this interference effect was larger in the 75% congruent condition than in the 0% condition (i.e., congruency proportion interacted with trial type), F(1, 83) = 34.14,MSE = 0.003.”

In Experiment 1's 75% congruent condition, response-time facilitation was significant within both the high-span and low-span groups.

“(facilitation was significant for both high-and low-span groups individually, both Fs > 35).”

The Stroop effect was significantly larger for participants in the 75% congruent condition than for those in the 0% congruent condition.

“this Stroop effect was significantly larger for participants in the 75% congruent condition than for those in the 0% condition (i.e., congruency proportion interacted with trial type), F(1, 83) = 48.23,MSE = 2,669.56.”

In Experiment 1, span differences in interference were robust only in the 75% congruent condition.

“Indeed, span differences in interference were robust in only the 75% congruent condition”

In Experiment 2, low spans had a higher overall error rate than high spans.

“low spans had a higher error rate overall than did high spans, F(1, 86) = 6.17,MSE = 0.004”

In Experiment 2's 75% congruent condition, low spans showed significantly larger response-time facilitation than high spans.

“low spans showed significantly larger facilitation effects than did high spans, F(1, 86) = 10.10,MSE = 1,008.31”

In Experiment 2, span groups differed significantly in response-time facilitation.

“Also as in Experiment 1, span groups differed significantly in response-time facilitation.”

In Experiment 2, low spans also showed larger response-time facilitation than high spans.

“Low spans also showed larger response-time facilitation effects than did high spans.”

In Experiment 1's 75% congruent condition, error interference was significant within both the high-span and low-span groups.

“(interference in the 75% condition was significant for both the high-and low-span groups individually, both Fs > 17).”

In Experiment 4, low spans committed more errors than high spans.

“Low spans com-mitted more errors than did high spans, F(1, 270) = 11.87,MSE = 0.007,”

In Experiments 3A and 3B, unlike Experiments 1 and 2, low spans demonstrated greater response-time interference than high spans.

“Thus, in contrast to Experiments 1 and 2, low spans in Experiments 3A and 3B demonstrated greater interference in latencies than did high spans.”

In Experiment 4, the 80% congruent task produced span differences in all manifestations of interference rather than selectively in errors or latencies depending on task order.

“the 80% congruent task produced span differences in all manifestations of interference, rather than selectively eliciting it in errors in high-congruency-first conditions and in latencies in high-congruency-last conditions.”

In Experiments 3A and 3B, no span differences appeared in error interference in any congruency condition.

“Unlike the previous experiments, then, no span differences were seen in error interference in any congruency condition, despite other effects being statistically detectable.”

In combined Experiments 3A and 3B, low spans appeared to show larger response-time interference than high spans in both the 0% and 75% congruency conditions.

“From inspection of the means, low spans appeared to show larger interference effects than did high spans in both the 0% and 75% proportion-congruency conditions.”

In Experiment 4, among error-prone participants, low spans had a larger error interference effect than high spans.

“even among this highly error-prone group, low spans had a larger error interference effect than did high spans (Ms = +.23 and +. 17, respectively), F(1, 67) = 4.40, MSE = 0.01.”

In Experiments 3A and 3B, there was a significant overall Stroop effect.

“a main effect of trial type, indicating a significant Stroop effect overall, F(1, 134) = 601.61,MSE = 3,414.53”

In the 0% and 20% congruent tasks, low spans showed somewhat larger response-time interference effects than high spans.

“in the 0% and 20% congruent tasks from all of our experiments, low spans showed somewhat larger response-time interference effects than did high spans”

In Experiments 3A and 3B, span differences in interference in the 75% task were evident in response times rather than errors.

“With respect to the 75% congruent task, span differences in interference were evident in Experiments 3A and 3B, but in response times rather than in errors.”

In Experiments 3A and 3B, span group interacted with trial type, indicating significant span differences in response-time interference.

“unlike the previous experiments, span group interacted with trial type, indicating significant span differences in response-time interference, F(1, 134) = 12.74, MSE = 3,414.53”

In Experiment 4, 80% congruent tasks produced larger interference effects than 20% congruent tasks overall.

“From inspection of the means, the 80% congruent tasks produced larger interference effects than did the 20% congruent tasks overall.”

In the 75% congruent task in Experiments 3A and 3B, where it followed the 0% task, span differences in interference appeared in latencies and not in errors.

“in the 75% congruent task in Experiments 3A and 3B, where it followed the 0% task, span differences in interference appeared in latencies, and not in errors.”

In Experiment 4, clean errors were committed much more quickly than ambiguous errors.

“clean errors were committed much more quickly than ambiguous errors, F(1, 57) = 58.46,MSE = 792,484.57.”

In Experiment 4, the effect of WM span on response latency was moderated by trial type, such that low spans showed greater response-time interference than high spans overall.

“was moderated by significant interactions of span with congruency proportion, F(1, 270) = 6.3 1, MSE = 4,642.86,and with trial type, F(1, 270) = 13.49,MSE = 3,517.11, the latter indicating greater interference for low spans than for high spans overall”

In Experiments 3A and 3B, high-span participants responded more quickly than low spans overall.

“High-span participants responded more quickly than did low spans overall, F(1, 134) = 6.58,MSE = 36,660.95”

Only Experiments 1 and 2, where the 75% congruent conditions did not follow 0% conditions, produced span differences in both error interference and response-time facilitation.

“Only Experiments 1 and 2, where the 75% congruent conditions did not follow 0% conditions, produced span differences in both error interference and response-time facilitation.”

In Experiment 4, the span difference in 80% congruent error interference appeared in both task orders.

“the span difference in 80% congruent errors appeared to be present in both task orders (i.e., for both 20%-first and 80%-first participants).”

In Experiment 4, span differences in error interference were larger in the 80% congruent condition than in the 20% congruent condition.

“span differences in error interference were larger in the 80% congruent condition than in the 20% congruent condition”

In Experiment 4, the real discrepancy with Experiments 1-3B was a significant span difference in error interference in the 80% congruent conditions regardless of task order.

“the real discrepancy it produced with Experiments 1-3B is the significant span difference in error interference in the 80% congruent conditions, regardless of task order.”

Presenting the 0% congruent task before the 75% congruent task produced span differences in response-time interference in both tasks, in both Experiments 3A and 3B.

“Presenting the 0% congruent task before the 75% congruent task produced span differences in response-time interference in both tasks, and it did so in both Experiments 3A and 3B.”

Low spans' error interference effects varied systematically by whether the 75% task followed the 0% task.

“low spans' effects varied systematically by whether the 75% task followed the 0% task.”

In Experiments 3A and 3B, response-time interference was significant for each span group.

“(interference was significant for each span group, with Fs > 3 10).”

Low spans committed many more errors on incongruent trials than high spans when congruent trials made up 75% or 80% of the Stroop trials.

“Low spans committed many more errors on incongruent trials than did high spans when congruent trials made up 75% or 80% of the Stroop trials”

In Experiments 3A and 3B, interference was greater in the 75% congruent condition than in the 0% condition.

“a Congruency Proportion X Trial Type interaction, F(1, 134) = 73.95,MSE = 1,009.21,indicating greater interference in the 75% than the 0% congruent condition.”

In Experiment 4, low spans' exaggerated vulnerability to error interference in 80% congruent conditions occurred whether the 80% task was completed first or after the 20% task.

“Thus, low spans' exaggerated vulnerability to error interference in 80% congruent conditions was present whether the 80% task was completed first or following the 20% task.”

In Experiments 3A and 3B, error rates were not sensitive to span differences in interference in either the 0% or 75% congruent condition.

“Inspection of the means suggests that, unlike response times, error rates were not sensitive to span differences in interference in either the 0% or 75% congruent conditions.”

In Experiment 4, response-time Stroop interference was larger in the 80% congruent condition than in the 20% congruent condition.

“Stroop interference effects were larger in the 80% congruent conditions than in the 20% congruent conditions, F(1, 270) = 670.28,MSE = 1,419.43.”

In the 0% and 20% congruent tasks, low spans did not show evidence of overt word reading.

“Here, low spans did not show evidence of overt word reading, suggesting that the presentation of mostly incongruent trials helped participants to keep track of the task goal of ignoring the word information.”

High spans' error interference effects in the 75% congruent task were strikingly stable and low across experiments regardless of task order.

“high spans' error interference effects in the 75% congruent task were strikingly stable (and low) across experiments, regardless of task order.”

In Experiment 4, low spans showed larger error interference than high spans only in the 80% congruent conditions.

“low spans showed larger interference effects than did high spans in only these 80% congruent conditions.”

In Experiment 4, error interference was larger in the 80% congruent condition than in the 20% congruent condition.

“it was larger for the 80% congruent condition than for the 20% condition, F(1, 270) = 132.15,MSE = 0.002.”

In Experiment 4, the subsequent 80% congruent task elicited greater error interference in low spans than in high spans.

“the subsequent 80% task elicited greater error interference in low spans than in high spans”

In the 75% congruent task data, response-time facilitation was positively correlated with error interference in Experiments 1, 2, 3A and 3B, with correlation coefficients between .34 and .45.

“Response-time facilitation correlated positively with error interference in Experiments 1, 2, 3A and 3B, with rs between .34 and .45.”

The locus of the WM-span difference in Stroop interference, whether in latency or accuracy, varied with congruency conditions and with the order of congruency conditions.

“the locus of the WM-span difference in Stroop interference-in latency or accuracy-varied with congruency conditions and the order of congruency conditions.”

Response-time facilitation showed no positive correlation with response-time interference, with correlation coefficients between -.15 and .25.

“we found no positive correlation between response-time facilitation and response-time interference (with rs between -. 15 and .25).”

• cites Lindsay & Jacoby, 1994Not checked yet
• cites Vanayan, 1992, cited in MacLeod, 1998Not checked yet

Error interference and latency interference did not significantly correlate with each other, with rs of -.02, .17 and .10 in Experiments 1-3B respectively.

“error and latency interference did not significantly correlate with each other (rs = -.02,.17,and .10 in Experiments 1-3B, respectively)”

In all five experiments, high-WM-span participants demonstrated less interference than low-span participants.

“In all five experiments, high-WM-span participants demonstrated less interference than did low-span participants.”

In the present experiments, WM-span differences were found in Stroop contexts that did not encourage goal neglect, such as 0% and 20% congruent conditions, where low spans showed greater response-time interference than high spans.

“in the present experiments we also found WM-span differences in Stroop contexts that did not encourage goal neglect, such as in 0% and 20% congruent conditions, where low spans showed greater response-time interference than did high spans”

• cites Miyake, Friedman, Emerson, Witzki, & Howerter, 2000Not checked yet

The proportion of congruent trials in the Stroop task drives WM span differences in error interference.

“Our discovery that the proportion of congruent trials in the Stroop task drives WM span differences in error interference”

• AssumptionThe authors state this as an assumption: Goal maintenance is less critical in 0% and 20% congruent contexts than in 75% and 80% congruent contexts, because repeated incongruent trials reinforce the ignore-the-word goal enough to aid goal-directed responding.

Low-WM-span individuals performed worse than high spans on Stroop tasks that present conflict between habit and goal.

“low-WM-span individuals demonstrated worse performance than high spans in Stroop tasks that present conflict between habit and goal.”

• cites Conway et al., 2001Not checked yet
• cites Kane et al., 2001Not checked yet

Studies that failed to link Stroop interference to intelligence or PFC generally minimized WM demands by presenting only incongruent trials in a continuous sequence.

“By our reading of the relevant literatures, studies failing to link Stroop interference to intelligence or PFC generally minimized WM demands of the Stroop task by presenting only incongruent trials in a continuous sequence.”

Stroop interference is jointly determined by two mechanisms: goal maintenance and competition resolution.

“The data and a literature review suggest that Stroop interference is jointly determined by 2 mechanisms, goal maintenance and competition resolution”

The finding that WM capacity predicts Stroop performance indicates that executive control and goal maintenance are important for selective attention.

“indicating the importance of executive control and goal maintenance to selective attention.”

Low spans have more difficulty resisting the lure of a powerful orienting cue than do high spans.

“These findings indicate that low spans have more difficulty resisting the lure of a powerful orienting cue than do high spans.”

Simple memory ability alone cannot explain the strong association between WM span and complex cognition.

“Our view is that simple memory ability, alone, cannot account for the strong association between WM span and complex cognition.”

Maintaining a secondary memory load increases reflexive errors in the antisaccade task by over 50%.

“maintaining a secondary memory load in-creases reflexive errors in the antisaccade task by over 50% (Roberts, Hager, & Heron, 1994).”

• cites Roberts, Hager, & Heron, 1994Not checked yet

In the Stroop task, color naming is slower and less accurate when the color and word conflict than when the word is unrelated or matching.

“When the color and word are in conflict, such as the word RED appearing in blue, color naming is slower and less accurate than when the word is unrelated to the color (e.g., PIN, or XXX, in blue), or when the color and word match (e.g., BLUE in blue).”

• cites MacLeod, 1991Not checked yet

Stroop interference can be viewed as analogous to dichotic-listening and antisaccade effects, representing failure of a goal to control behavior against a conflicting habitual response.

“Stroop interference may be viewed as analogous to the dichotic-listening and antisaccade effects described above, insofar as it represents the failure of a goal (-ignore the word‖) to control behavior in the face of a conflicting habitual response (reading the word).”

Information maintenance in the face of interference is the critical function of WM capacity.

“Information maintenance in the face of interference is the critical function of WM capacity”

WM and STM span tasks are strongly correlated at the latent-construct level.

“WM and STM span tasks were strongly correlated with each other at the latentconstruct level, reflecting that both involved the retention of unrelated words.”

• cites Conway, Cowan, Bunting, Therriault, & Minkoff, 2002Not checked yet
• cites Engle, Tuholski, Laughlin, & Conway, 1999Not checked yet

Blocked presentations of incongruent trials may minimize WM involvement because task demands remain consistent across trials, making it easier to keep the goal in mind.

“Such blocked presentations of incongruent trials may minimize WM involvement because the task demands remain consistent across trials, making it easier for participants to keep the goal in mind.”

A mixed-trial Stroop task including congruent stimuli places more premium on keeping unusual task requirements active in WM than a blocked-trial task with only incongruent stimuli.

“A mixed-trial Stroop task that includes congruent stimuli puts more of a premium on keeping the unusual task requirements active and accessible in WM than does a blocked-trial task that presents only incongruent stimuli.”

The goal-maintenance mechanism is partly responsible for producing Stroop facilitation effects sometimes observed on congruent trials compared with neutral trials.

“Our data also imply that the goal-maintenance mechanism is partly responsible for producing the Stroop -facilitation‖ effects that are sometimes observed on congruent trials compared with neutral trials”

• cites MacLeod, 1998 sourceNot checked yet
• cites MacLeod & MacDonald, 2000aNot checked yet
• cites MacLeod & MacDonald, 2000bNot checked yet

• AssumptionRests on an assumption: Participants complied with the instruction to ignore the word on every trial and did not look away or squint during the Stroop task.

In Experiment 1's high-congruency context, low spans made more errors than high spans on rare trials where the goal to ignore the word conflicted with the habit to read it.

“Clearly, in a high-congruency context, which allowed for goal neglect without frequent cost, low spans made more errors than did high spans on those rare trials where the goal (-ignore the word‖) and habit (-read the word‖) came in direct conflict.”

• AssumptionRests on an assumption: Goal neglect in the Stroop task leads to overt word reading.

Research outside individual differences has shown that the magnitude of Stroop and Stroop-like interference increases with the proportion of congruent trials in the design.

“Research outside of the individual-differences do-main has indeed shown that the magnitude of Stroop and Stroop-like interference increases with the proportion of congruent trials in the design (e.g., Logan & Zbrodoff, 1979, 1998).”

• cites Logan & Zbrodoff, 1979Not checked yet
• cites Logan & Zbrodoff, 1998Not checked yet

In measures and contexts more sensitive to goal maintenance and neglect, older adults, schizophrenics, and Alzheimer's patients show substantial increases in errors or in the skew of the response-time distribution on incongruent trials.

“However, in measures and contexts more sensitive to goal maintenance and neglect, these participant groups, along with Alzheimer's patients (Spieler et al., 1996), show substantial increases either in errors or in the skew of the response-time distribution on incongruent trials.”

• cites Spieler et al., 1996Not checked yet

In Experiment 1's 75% congruent condition, low spans were less likely than high spans to ignore word information, as indicated by their larger facilitation.

“As indicated by facilitation, then, low spans appeared less likely than high spans to ignore word information during the 75% congruent task.”

A Stroop context with many congruent trials allows successful responding on most trials even if participants neglect the goal and respond habitually by reading the word.

“A Stroop context including many congruent trials will allow successful responding on most trials even if participants completely neglect the goal and respond habitually by reading the word.”

In Experiment 1, the 75% congruent task promoted goal neglect, especially in low-WM-span individuals.

“The 75% congruent task, in contrast, appears to have promoted goal neglect, particularly for low-WM-span individuals.”

• AssumptionRests on an assumption: Goal neglect in the Stroop task leads to overt word reading.

The robust span differences in interference in Experiment 1's 75% condition suggest span differences in goal neglect.

“suggesting span differences in goal neglect.”

• AssumptionRests on an assumption: Goal neglect in the Stroop task leads to overt word reading.

The significant span difference in error interference in Experiment 2 replicates the corresponding result from Experiment 1.

“The significant span difference in interference clearly replicates that from Experiment 1.”

Low spans have a relatively enduring deficit in the competition-resolution mechanism that is independent of goal maintenance.

“we propose that low spans have a relatively enduring deficit in the competition-resolution mechanism, independent of goal maintenance.”

A second mechanism of Stroop interference is a transient failure of goal maintenance, reflected most obviously in accuracy and prevailing in contexts that maximize the need for goal maintenance.

“The other, additional mechanism is a more transient failure of goal maintenance, reflected most obviously in accuracy and prevailing in contexts that maximize the need for goal maintenance.”

In Experiment 4, participants who committed goal-neglect errors did so quickly regardless of WM capacity.

“Thus, participants who committed goal-neglect errors did so quickly, regardless of WM capacity.”

WM differences in error interference in 75% and 80% congruent tasks not preceded by 0% tasks resulted largely from momentary failures of goal maintenance in low spans.

“the WM differences in error interference in 75% and 80% congruent tasks that were not preceded by 0% tasks resulted largely from momentary failures of goal maintenance in low spans.”

One mechanism of Stroop interference is a consistently time-consuming process of competition resolution, reflected primarily in latencies and prevailing in contexts that minimize the need for goal maintenance.

“One mechanism is a rather consistently time-consuming process of competition resolution, which is reflected primarily in latencies and prevailing in contexts that minimize the need for goal maintenance.”

The lack of a span difference in error interference for a 75% task following a 0% task suggests that practice in a 0% congruent context reduced the subsequent probability or severity of goal neglect in low-span participants.

“The lack of a span difference in error interference for a 75% congruent task that followed a 0% task (despite significant interference overall) suggests that practice in a 0% congruent context reduced the subsequent probability or severity of goal neglect in low-span participants.”

• AssumptionThe authors state this as an assumption: Some degree of goal maintenance is necessary even in the 0% congruent Stroop task.
• AssumptionRests on an assumption: Goal neglect in the Stroop task leads to overt word reading.

The specific patterns of span effects suggest that Stroop interference has two causes.

“we suggest that the specific patterns of span effects observed here suggest that Stroop interference has two causes, as we described in the introduction.”

At least with respect to accuracy, low spans were more affected by set transfer than high spans, responding in the 75% task as they had in the previous 0% task.

“At least with respect to accuracy, then, low spans were more affected by set transfer than were high spans, responding in the 75% task as they had in the previous 0% task.”

Congruency-task order appears to have a robust effect on the pattern of span differences in Stroop interference.

“Thus, congruency-task order appears to have a robust effect on the pattern of span differences in Stroop interference.”

Span differences in error versus latency interference were strongly influenced by the current and previous task contexts.

“Our results clearly demonstrate that span differences in error versus latency interference were strongly influenced by the current and previous task contexts.”

Experiments 1, 2, and 4 demonstrated that WM span is related to goal maintenance and neglect in the Stroop task.

“Experiments 1, 2, and 4 demonstrated in a straightforward manner that WM span is related to goal maintenance and neglect in the Stroop task.”

In Experiment 4, the 80% task that followed the 20% task produced findings indicative of overt word reading in low-span participants.

“in Experiment 4, the 80% task that followed the 20% task also produced findings indicative of overt word reading (i.e., failed goal maintenance and subsequent errors) in low-span participants.”

Without external goal reinforcement from frequent incongruent trials, low spans were more likely to lose access to the goal and respond by reading the word in error.

“Without the external goal reinforcement provided by frequent incongruent trials, low spans were more likely to lose access to the goal and respond according to habit, by reading the word in error.”

Experiments 1 through 3B indicate that the locus of WM-span-related differences in Stroop interference depends on both the congruency proportions within the task and the order in which those proportions are encountered.

“Thus, Experiments 1-3B indicated that the locus of WM-span-related differences in Stroop interference depends not only on the congruency proportions within the task, but also on the order in which congruency proportions are encountered.”

When the prior 0% congruent context reinforced the task goal by repeatedly presenting incongruent trials, it appeared to establish a task set that transferred to the subsequent 75% congruent task.

“when the prior 0% congruent context reinforced the goal of the task by repeatedly presenting incongruent trials, it appeared to establish a task set that transferred to the subsequent 75% congruent task.”

The present data support MacLeod's inadvertent reading hypothesis.

“We believe that our data support MacLeod's (1998;MacLeod & MacDonald, 2000a, 2000b) -inadvertent reading‖ hypothesis.”

• cites MacLeod's (1998;MacLeod & MacDonald, 2000a, 2000b)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Active maintenance and distractor blocking are interdependent executive-attention processes that are central to WM capacity.

“active maintenance and distractor blocking are interdependent executive-attention processes that are central to WM capacity.”

In the absence of interference, task-relevant information or goals can be retrieved on demand from inactive long-term memory.

“In the absence of interference, task-relevant information or goals can be retrieved on demand from inactive long-term memory.”

The absence of a positive correlation between facilitation and interference is more likely to reflect the statistical artifact of computing facilitation and interference effects from the same neutral baseline.

“it is more likely to reflect the statistical artifact of calculating facilitation and interference effects from the same neutral baseline”

• cites Watson, Brown, & Stott, 2001Not checked yet

There are important interactions between WM capacity and task context in mediating Stroop interference.

“our findings, along with Long and Prat's, clearly indicate that there are important interactions between WM capacity and task context in mediating Stroop interference.”

Stroop facilitation is at least partly dependent on the goal-neglect mechanism that largely contributes to interference in 75% congruent Stroop tasks and in other conditions that put a premium on effective goal maintenance.

“However, facilitation is, in fact, at least partly dependent on the goal-neglect mechanism that largely contributes to interference in 75% congruent Stroop tasks (and in other conditions that put a premium on effective goal maintenance).”

In the Long and Prat study, low proportions of incongruent trials accompanied high proportions of neutral trials rather than congruent trials.

“in the Long and Prat study, low proportions of incongruent trials accompanied high proportions of neutral trials (XXXXX) rather than congruent trials.”

• cites Long & Prat, 2002Not checked yet

In the presence of interference, task-irrelevant information and action schemas are likely to be retrieved, resulting in errors or slowed responding.

“In the presence of interference, however, task-irrelevant in-formation and action schemas are likely to be retrieved, resulting in errors or slowed responding”

• cites Norman, 1981Not checked yet
• cites Reason, 1990 sourceNot checked yet

A context high in neutral trials should not induce goal neglect, because word reading is not a possible response on those trials.

“A context that is high in neutral trials should not induce goal neglect because word reading is not a possible response on these trials”

Research with patient groups such as those with schizophrenia and Alzheimer's disease indicates exaggerated interference effects primarily in errors rather than response times.

“research with patient groups, such as those with schizophrenia (e.g., Perlstein et al., 1998) and Alzheimer's disease (Spieler et al., 1996), indicate exaggerated interference effects primarily in errors, as opposed to response times.”

• cites Perlstein et al., 1998Not checked yet
• cites Spieler et al., 1996Not checked yet

The two indices of interference may reflect different mechanisms.

“further suggesting that the two indices of interference may reflect different mechanisms.”

Tasks that include even a small proportion of congruent trials do not induce a task set strong enough to transfer to a subsequent high-congruency task.

“tasks that include even a small proportion of congruent trials do not induce a task set that is strong enough to transfer to a subsequent high-congruency task.”

The mechanism responsible for the broad predictive power of WM span tasks is a domain-general executive-attention mechanism that maintains stimulus representations, action plans and goals in the face of interference.

“the mechanism responsible for the broad predictive power of WM span tasks is a domain-general, executive-attention mechanism that allows for the maintenance of stimulus representations, action plans, and goals in the face of interference”

• cites Engle, 2001 sourceNot checked yet
• cites Engle, Kane, & Tuholski, 1999 sourceNot checked yet

WM capacity, as defined by complex span measures, is a valid predictor of attention control.

“These findings are consistent with the idea that WM capacity, as defined by complex span measures, is a valid predictor of attention control.”

Stroop facilitation may be independent of the competition-resolution process that largely contributes to interference in 0% congruent tasks and in other conditions that minimize the need for goal maintenance.

“Facilitation may be independent of the competition-resolution process that largely contributes to interference in 0% congruent tasks and in other conditions that minimize the need for goal maintenance.”

The present data support a role for goal neglect in determining the Stroop facilitation effect.

“our data clearly do support a role for goal neglect in determining the effect.”

Long and Prat (2002) found that WM span is differentially related to Stroop interference according to incongruency-proportion contexts.

“Long and Prat (2002) found that WM span is differentially related to Stroop interference according to incongruency-proportion contexts.”

• cites Long & Prat, 2002Not checked yet

The discrepancy between the present findings and Long and Prat's is almost certainly due to their use of neutral rather than congruent trials.

“so this is almost certainly the source of the discrepancy between our findings.”

WM capacity may be linked to additional attentional capabilities that are not obviously involved in active goal maintenance in the face of interference.

“indicates that WM capacity may be linked to additional attentional capabilities that are not obviously involved in active goal maintenance in the face of interference.”

Low spans showed greater response-time facilitation than high spans only when they were also more likely to neglect the goal and read words overtly.

“low spans only showed greater response-time facilitation than high spans when they were also more likely to neglect the goal and read words overtly.”

The 0% congruent task appeared to elicit a task set that transferred to the subsequent 75% task.

“the 0% congruent task appeared to elicit a task set that transferred to the subsequent 75% task.”