Learning and Motivation 56 (2016) 53–64

Contents lists available at ScienceDirect

Learning and Motivation journal homepage: www.elsevier.com/locate/l&m

Stroop-like interference in a match-to-sample task: Further evidence for semantic competition? Marshall L. Green, Lawrence Locker Jr., Ty W. Boyer, Bradley R. Sturz ∗ Georgia Southern University, United States

a r t i c l e

i n f o

Article history: Received 21 July 2016 Received in revised form 12 September 2016 Accepted 14 September 2016 Keywords: Stroop Match-to-sample Semantic competition Response competition

a b s t r a c t Two explanations have emerged to account for the interference of word reading on color naming observed in the canonical Stroop task. Semantic competition suggests that interference results from competing semantic processes associated with the word and color dimensions of the stimulus. Response competition suggests that interference results from competition in articulating the word versus the color dimension. Recently, Sturz et al. (2013) attempted to reproduce a Stroop-like phenomenon within the context of a delayed matchto-sample (DMTS) task. Importantly, this task format provided an opportunity to isolate semantic versus response competition, through the manipulation of the congruence of the bi-dimensional samples’ font color and word meaning and the relatedness of the foil to the irrelevant sample dimension. Findings indicated that incongruent samples produced Stroop-like interference, regardless of whether the foil was related with the irrelevant sample dimension or not, which was interpreted as support for semantic competition within the DMTS task. The present experiments further examine Stroop-like interference in the MTS task by manipulating the stimulus onset asynchrony (SOA). In Experiment 1, we presented the Stroop sample and response options sequentially, but without a retention interval, and in Experiment 2, we presented the sample and response options simultaneously. The results indicated increased reaction times on incongruent trials, independent of whether or not the foil was related to the irrelevant sample dimension. An asymmetrical Stroop-like pattern of interference, where the sample word interfered with color matching but not the reverse, was only observed in Experiment 2. Collectively, these results suggest that empirical and theoretical findings obtained in the traditional Stroop task may generalize to the DMTS task. © 2016 Elsevier Inc. All rights reserved.

1. Introduction In 1935, J.R. Stroop designed what is now widely known as the Stroop task. The task consisted of participants naming the ink colors of a list of color-words, and participants were much slower to respond when the words differed from the ink color they represented. Stroop’s (1935) theoretical explanation of this effect was that associations between the word stimuli and the reading response were stronger than the associations between color stimuli and the naming response. At the time, Stroop suggested that reading a word was a strong association that overrode the relatively weak “to name” association.

∗ Corresponding author at: Department of Psychology, Georgia Southern University, P.O. Box 8041, Statesboro, GA 30460, United States. E-mail address: [email protected] (B.R. Sturz). http://dx.doi.org/10.1016/j.lmot.2016.09.003 0023-9690/© 2016 Elsevier Inc. All rights reserved.

54

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

Since this initial interpretation was offered, the relative speed-of-processing account has been suggested to explain this “Stroop effect.” Specifically, it is posited that word reading is faster than color naming. The advantage of this account is that it assumes a continuum of processing speed such that the faster process of reading provides a potential response that competes with, and thereby slows the color-naming process (MacLeod, 1991, 1992; Posner & Snyder, 1975). This relative speed of processing account has been examined through manipulations of stimulus onset asynchrony (SOA) – separating the bi-dimensional stimulus into separate word and color patch stimuli and systematically manipulating the time between presentation of the relevant color patch and the irrelevant word. The logic is that if the color naming process indeed takes longer than the word reading process, then an SOA between the presentation of the color and the color-word of sufficient length should allow for completion of the “naming” processes and eliminate slowed responding. Importantly, this does decrease the Stroop effect (see Glaser & Glaser, 1982; MacLeod & Dunbar, 1988), which has given rise to an automaticity and attention account that emphasizes that reading requires less attention than naming a color (MacLeod, 1991). Although the automaticity account is capable of explaining results from SOA manipulations, some argue that the effects are more parsimoniously characterized as contextually controlled rather than automatic (Besner, Stolz, and Boutilier, 1997; see also Kahneman & Henik, 1981; MacLeod & Dunbar, 1988). An alternative explanation is that incongruent Stroop stimuli activate semantic representations of both color and word dimensions, producing semantic competition between these dimensions prior to response selection (Augustinova & Ferrand, 2012; Augustinova, Flaudias, & Ferrand, 2010; Klein, 1964; Schmidt & Cheesman, 2005; Luo, 1999). That is, one component of the Stroop effect results from attending to both the color and word dimensions that then activate competing semantic representations – generating a response along one dimension requires the active suppression of the other dimension. This additional suppression slows processing and thus responding. When both of the dimensions represent the same semantic code (i.e., when the color and word are congruent), responding is not affected because suppression of the irrelevant semantic code is not required. In contrast, response competition posits that incongruent dimensions of Stroop stimuli activate response units that produce interference at the point of response output (Luo, 1999; De Houwer, 2003). This explanation suggests that one component of the Stroop effect results from a bi-dimensional Stroop stimulus containing a ‘respond’ dimension (e.g., name the color) as well as a ‘do not respond’ dimension (e.g., read the word). Thus, the response can only be made by increasing attention to the ‘respond’ over the ‘do not respond’ dimension (MacLeod, Chiappe, & Fox, 2002). By definition, response competition occurs later in processing and places Stroop interference at the level of executive control. Recently, Sturz, Green, Locker, and Boyer (2013) developed a delayed match-to-sample (DMTS) Stroop-like task in an attempt to isolate the effects of the activation of semantic codes during processing and response options at response selection. The task involved viewing one of three sample types: (1) a congruent sample (e.g., a color word such as “red” in red font color), (2) an incongruent sample (e.g., a color word such as “red” in blue font color), or (3) a baseline sample (e.g., a color word such as “red” in black font color). The presentation of the sample was followed by a brief delay. After the delay, two unidimensional response option stimuli were presented: (1) a matching target, and (2) a non-matching foil. Using this approach, the task afforded a unique opportunity to examine responses to both the word and font color dimensions through manipulation of the response options presented on a given trial. Specifically, response options were either two words (e.g., “red” and “blue” presented in black font) or two color patches (see Fig. 1 for example trials). Participants were instructed to select the word that matched the preceding sample word dimension if the options were words (e.g., select the word “red” if the preceding sample had been the word “red,” regardless of font color) or to select the color patch that matched the preceding sample font color if the options were two color patches. This task also provided an opportunity to systematically manipulate the relatedness of the foil response option to the irrelevant sample dimension. For example, on a trial where the sample was the word “red” in blue font, if two color patch options appeared, the match would be a blue color patch and the foil could be related to the irrelevant word dimension (i.e., a red color patch). Alternatively, the foil could be unrelated to the sample (i.e., a yellow color patch). Similarly, if two word options appeared, then the match would be the word “red” and the foil could be related to the irrelevant color dimension (i.e., the word “blue”), or could be unrelated (i.e., the word “yellow”). It is important to emphasize that in the DMTS Stroop task the relevant dimension, on which participants based their responses, was ambiguous prior to response option onset. Consequently, on incongruent sample trials, participants were required to attend to, and retain, both sample dimensions for the duration of the retention interval, and, therefore, the task demands differed from typical Stroop tasks, which require that the participant attend to only one dimension and explicitly disregard the other dimension (Besner et al., 1997; Cohen, Dunbar, & McClelland; 1990). Sturz et al. (2013) found that word targets were responded to more slowly than color targets, and, notably, that, for both color and word response options, reaction times were slower for the incongruent sample condition than congruent or baseline sample conditions, regardless of whether or not the foil was related to the irrelevant sample dimension. There was, however, a decrement in accuracy for incongruent conditions in which the foil was related to the irrelevant dimension for both word and color targets. Although a traditional Stroop effect has been suggested to result from a combination of semantic and response competition (e.g., Augustinova & Ferrand, 2012; Augustinova et al., 2010; De Houwer, 2003; MacLeod et al., 2002), Sturz et al. (2013) explained these findings in terms of semantic competition. As noted, the task required participants to attend to both sample dimensions on incongruent trials because the eventual response dimension remained ambiguous until the options appeared, thus requiring activation of semantic codes for both dimensions. Furthermore, on incongruent sample trials, the dimension rendered irrelevant by the appearance of the options must then be suppressed. By contrast, the lack of an effect of whether the foil was related or unrelated with the irrelevant sample dimension was taken as a failure to find evidence of response

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

55

Fig. 1. Sample Trial Types and trial structures for Testing trials for the Delayed Match-to-Sample task. One sample trial is illustrated for Word Targets (a.) and Color Targets (b.) for Congruent, Incongruent – Unrelated Foil, and Incongruent – Related Foil trial types. For illustrative purposes, all correct matches are shown as the left target even though correct target location was balanced (see text for details). Please refer to online version for a full color version of this figure.

competition. That is, if response conflict had arisen, then the RTs for related foils (e.g., a red color patch or the word “blue” target following a sample with “red” in blue font) should have exceeded those for unrelated foils (e.g., “yellow” or a yellow color patch target following a “red” in blue font sample), yet, RTs for these two incongruent sample conditions were not significantly different. The sequential presentation format Sturz et al. (2013) utilized, with a 5-s and 10-s stimulus onset asynchrony (SOA) between the sample stimulus offset and response options onset may have contributed to the observed effects. Essentially, this delay may have served as a period for semantically mapping both the word and color dimensions of incongruent bidimensional stimuli, effectively enhancing semantic interference effects and offsetting response competition effects. Thus, in the present experiments, we manipulated the extent to which a foil can compete with the target for response selection during sequential (Experiment 1) or simultaneous (Experiment 2) presentation of sample and response options. We predict, consistent with Sturz et al. (2013), that encoding word and color dimensions of the sample stimuli prior to knowing the relevancy of either dimension (i.e., as in the sequential format of Experiment 1) should produce semantic competition and result in increased RTs and error rates on trials in which the sample dimensions are incongruent (e.g., “red” in blue font), and that this will apply to both stimulus dimensions (i.e., matching sample-target words and colors). In

56

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

contrast, presenting samples in the presence of the response options (i.e., as in the simultaneous format of Experiment 2), is more analogous to the traditional Stroop task in which attention may be biased toward word reading. Therefore, we predict that knowing the relevant dimension at the trial onset should produce semantic competition and result in increased RTs and error rates on trials in which the sample dimensions are incongruent (e.g., “red” in blue font), but only for color targets. Finally, if the delay implemented by the Sturz et al. (2013) SOA manipulation dampened the ability to observe response competition, then eliminating this SOA, particularly with the simultaneous presentation format of Experiment 2, should increase RTs when the foil is related to the irrelevant sample dimension, compared to when the foil is unrelated with the irrelevant sample dimension. The directives for both Experiment 1 and 2 were to match the target to the sample, and the response option dimension informed the participant whether the task was to match word-to-word or color-to-color. In Experiment 1, both the word and color of the sample needed to be encoded and retained until response option onset to select the matching target. In Experiment 2, where the sample and response options were presented simultaneously, the sample must be attended to, but not necessarily retained, in order to match-to-sample. 2. Experiment 1 2.1. Method 2.1.1. Participants Twenty undergraduate students served as participants (6 males, 14 females). Participants had normal or corrected-tonormal vision. Participants were required to be age 18 or older. Compensation for participation consisted of extra course credit. The experimental protocol was approved by the Institutional Review Board of Georgia Southern University. 2.1.2. Apparatus A computerized match to sample (MTS) task was presented with a 22-in. flat screen liquid crystal display (LCD) monitor (1680 × 1050 pixels). Responses were made by depressing the “c” (left index finger) and “m” (right index finger) keys on a standard keyboard. Experimental events were controlled and recorded using E-prime (Psychology Software Tools, Inc., www.pstnet.com). Up to five participants completed the task concurrently and were separated by a partition within the experimental room. 2.1.3. Stimuli The two stimulus types were colors and words and were identical to those used by Sturz et al. (2013). Color stimuli were blue, red, and yellow color patches represented as a 410 × 410 pixel filled diamond subtending a 9.6◦ visual angle horizontally and vertically. Word stimuli were “blue”, “red”, and “yellow” and were presented in black, blue, red, or yellow font color depending on trial type. Word stimuli were presented in bold 48 point Courier New font and were 149 (“blue”), 112 (“red”), or 228 (“yellow”) pixels in width, subtending 2.6◦ visual angle horizontally, and 40 (“blue” and “red”) or 52 (“yellow”) pixels in height, subtending 0.9◦ visual angle vertically. All stimuli were presented on a white background. Samples were presented in the center of the screen 25% down from its top edge. Targets were presented on opposite sides of the screen, 50% of screen width apart, and 25% up from its bottom edge (see Fig. 1). 2.1.4. Procedure Participants were instructed that they were completing a matching test in which they would be presented with either colored shapes or words. They were informed to press the “c” key (left hand) if the matching word/color was on the left, and to press the “m” key (right hand) if the matching word/color was on the right. Experiment 1 consisted of 120 trials for each participant and was modeled after the design of Sturz et al. (2013). First, training trials were provided in order to familiarize participants with the MTS task. There were 24 training trials composed of two 12-trial blocks. One block consisted of 12 unique color training trials in which the sample presented was a diamond shaped color patch and targets were two diamond shaped color patches (one match and one foil). The second block consisted of 12 unique word training trials in which the sample was presented in black font only and the response options were presented in black font only (one match and one foil). The order of training block presentation was counterbalanced across participants. Testing consisted of 96 trials composed of twelve 8-trial blocks (24 Baseline, 24 Congruent trials, 24 IncongruentUnrelated trials, and 24 Incongruent-Related trials – see Fig. 1). Baseline trials were identical to the training trials mentioned above (i.e., samples were either words presented in black font or a color patches and were matched to an identical target). On Congruent trials, the Stroop sample was one of the three word stimuli presented in its corresponding font color and the response options were the match and a foil. IncongruentUnrelated Foil trials consisted of an incongruent Stroop sample (i.e. color-word in non-corresponding font color) and the response options were the match (i.e. word dimension match or color dimension match) and a foil which was unrelated to the irrelevant dimension. For example, if the sample was the word “blue” written in red font color, then on IncongruentUnrelated Foil word target trials the foil would be the word “yellow”. Incongruent-Related Foil trials also consisted of an incongruent Stroop sample, but the foil corresponded to the irrelevant dimension. For example, if the sample was the word

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

57

Fig. 2. Mean response time on correct trials during Testing (in milliseconds) plotted by Trial Type for Color Targets (filled bars) and Word Targets (unfilled bars) for Experiment 1. Error bars represent standard errors of the means.

“blue” written in red font color, then the match would be the word “blue” and the foil would be the word “red” (see Fig. 1 for example trials). One color target trial and one word target trial was presented for each of the four trial types within each block of eight trials, in random order. The left/right location of the match and foil was counterbalanced, for a unique combination of each trial type being presented once without replacement, resulting in a total of 96 test trials. On all trials, the sample was presented for 1000 ms, after which it was removed and followed immediately by the response options for 1500 ms, followed by feedback provided in the form of a green checkmark for correct responses, a red X-mark for incorrect responses, and “no response” for failing to make a response (see Fig. 1), during a 500 ms inter-trial interval (ITI). 2.2. Results 2.2.1. Response times Error trials and trials in which participants failed to respond were excluded from the RT analyses, resulting in the elimination of 37 of the total 1920 trials (1.9% total). Fig. 2 shows the mean RT on correct trials plotted by Trial Type for Color and Word Targets. A two-way repeated measures analysis of variance (ANOVA) with Target Type (color targets, word targets) and Trial Type (baseline, congruent, incongruent-unrelated foil, and incongruent-related foil) as factors revealed a main effect of Target Type, F(1, 19) = 71.75, p < 0.001, ␩2 p = 0.79, and a main effect of Trial Type, F(3, 57) = 28.66, p < 0.001, ␩2 p = 0.60. The Target Type x Trial Type interaction was not significant, F(3,57) = 0.17, p = 0.91, ␩2 p = 0.01. Participants were faster to respond to Color Targets compared to Word Targets. Fisher’s Least Significant Difference (LSD) post hoc paired samples t-tests conducted on the Trial Type factor revealed that response times for Baseline trials were significantly lower than Congruent trials [t(19) = 2.64, p = 0.02, d = 0.59], IncongruentUnrelated Foil Trials [t(19) = 7.72, p < 0.001, d = 1.72], and Incongruent-Related Foil Trials [t(19) = 4.92, p < 0.001, d = 1.07]. Congruent trials were significantly faster than Incongruent-Unrelated Foil [t(19) = 7.18, p < 0.001, d = 1.60], as well as Incongruent-Related Foil trials [t(19) = 5.09, p < 0.001, d = 1.13]. Incongruent-Unrelated Foil and Incongruent-Related Foil trials were not significantly different from each other, t(19) = 1.15, p = 0.26, d = 0.25. 2.2.2. Error rates Trials in which participants failed to respond (15 of the total 1920 trials, 0.8% total) were excluded. Fig. 3 shows the mean error rate plotted by Trial Type for Color and Word Targets. A two-way repeated measures ANOVA with Target Type (color targets, word targets) and Trial Type (baseline, congruent, incongruent-unrelated foil, and incongruent-related foil)

58

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

Fig. 3. Mean error rates during Testing plotted by Trial Type for Color Targets (filled bars) and Word Targets (unfilled bars) for Experiment 1. Error bars represent standard errors of the means.

as factors revealed a main effect of Trial Type, F(3,57) = 75.11, p < 0.001, ␩2 p = 0.80. There was no main effect of Target Type, F(1,19) = 1.87, p = 0.19, ␩2 p = 0.09. The Trial Type x Target Type interaction was significant, F(3,57) = 7.90, p < 0.001, ␩2 p = 0.29. To isolate the source of the Target Type x Trial Type interaction, separate one-way repeated measures ANOVAs with Trial Type as a factor were conducted for each Target Type. For Color Targets, there was a main effect of Trial Type, F(3, 57) = 66.67, p < 0.001, ␩2 p = 0.78. LSD post hoc tests revealed that error rates for Baseline trials did not significantly differ from Congruent trials, t(19) = 1.00, p = 0.33, d = 0.22. Error rates for Baseline trials were significantly lower than Incongruent-Unrelated Foil trials [t(19) = 3.607, p = 0.002, d = 0.80], as well as Incongruent-Unrelated Foil trials [t(19) = 8.59, p < 0.001, d = 1.92]. Error rates for Congruent trials were significantly lower than Incongruent-Related Foil trials [t(19) = 3.58, p = 0.002, d = 0.80], as well as Incongruent-Related Foil trials [t(19) = 8.54, p < 0.001, d = 1.91]. Error rates for Incongruent-Unrelated Foil trials were significantly lower than Incongruent-Related Foil trials, t(19) = 7.94, p < 0.001, d = 1.77. For Word targets, there was a main effect of Trial Type, F(3, 57) = 19.02, p < 0.05, ␩2 p = 0.50. LSD post hoc tests revealed that error rates for Baseline trials were not significantly different than Congruent trials [t(19) = 0.25, p = 0.80, d = 0.05] or Incongruent-Unrelated Foil trials [t(19) = 0.88, p = 0.39, d = 0.19]. Baseline trials had significantly lower error rates than Incongruent-Related Foil Trials, t(19) = 4.75, p < 0.001, d = 1.06. Error rates for Congruent trials did not significantly differ from those on Incongruent-Unrelated Foil trials, t(19) = 0.91, p = 0.38, d = 0.20. Error rates for Congruent trials were significantly lower than Incongruent-Related Foil trials [t(19) = 4.75, p < 0.001, d = 1.06], and Incongruent-Unrelated Foil trials had significantly lower error rates than Incongruent-Related Foil trials [t(19) = 5.15, p < 0.001, d = 1.15]. One-sample t-tests revealed that mean error rate was significantly less than chance (0.5) for all Trial Types and Target Types, ts(19) > 2.35, ps < 0.05. 2.3. Discussion The RT analyses indicated that responses for Congruent trials took longer than Baseline trials, and that both incongruent trial types, with unrelated and related foils, took longer than Baseline and Congruent trial types. This pattern of interference suggests that both sample dimensions were retained on incongruent sample trials, and we suggest that one of these dimensions therefore needed to be suppressed in order to identify the correct match, which we interpret as consistent with the semantic competition hypothesis. Critically, however, RTs on Incongruent-Unrelated Foil and Incongruent-Related Foil trials were not significantly different from one another, which fails to support the hypothesis of competition emerging at the point of response, even under the present conditions of a 0-s sample-target SOA. The analysis of error rates indicated a decrement in performance only when the sample was incongruent and the foil response option was related to the irrelevant dimension of the sample. Although, taken at face value, this finding seems

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

59

relatively at odds with the RT data, the incongruent-related foil trial type is the only instance where the foil is a viable, albeit incorrect, response option. If a participant were to match an irrelevant word to the color foil, it would technically be an error; however, the error would be that the wrong task was performed (i.e., match word to the name of the color) rather than evidence of a distractor response code competing for selection. The Stroop task and Stroop-like paradigms typically reveal asymmetrical effects, such that words interfere with the identification of colors, but colors do not interfere with the reading of words (Sabri, Melara, & Algom, 2001). Importantly, the size of the effect appears to be related to response modality such that effect sizes with verbal responses are larger than those with key press responses (see, Blais & Besner, 2006). Interestingly, neither Sturz et al. (2013) nor this experiment exhibited this asymmetrical pattern, even with the elimination of the retention interval between the sample and response options. One possibility is that the removal of the sample with the presentation of the response options modulates the bias in selectively attending to words over colors of incongruent sample stimuli. To further test this possibility, in Experiment 2 we presented the sample and response options simultaneously rather than sequentially. Our rationale is that if the sample and response options are presented simultaneously, the dimension of the response options dictates the relevant sample dimension and directs attention towards that dimension, and an automatic bias of word reading may emerge, producing an asymmetrical failure of selective attention for word versus color matching. 3. Experiment 2 In Experiment 2, the sample and response options were presented simultaneously for the duration of a trial to probe the effects of matching a target to a bi-dimensional stimulus. In Experiment 1, the relevant sample dimension was not known for any given trial until after the sample was removed and the response options were presented, therefore requiring encoding and retention of both the color and word dimensions of the sample. In Experiment 2, by contrast, the sample and response options appeared simultaneously, and, therefore, the relevant dimension was indicated at the onset of each trial, which should allow participants to immediately determine which sample dimension is relevant. Under the assumption of a reading bias in the presence of word stimuli, word response options should bias attention towards reading, and matching should occur rapidly and accurately with limited interference on word-matching trials. In contrast, color response options will not produce an automatic identification bias, and, thus, color-matching trials should produce asymmetrical Stroop-like interference. 3.1. Methods 3.1.1. Participants Twenty undergraduate students (6 males, 14 females), who had not participated in Experiment 1, served as participants. Participants had normal or corrected-to-normal vision. Compensation for participation consisted of extra course credit. The experimental protocol was approved by the Institutional Review Board of Georgia Southern University. 3.1.2. Apparatus, stimuli, and procedure The apparatus, stimuli, and procedure for Experiment 2 were identical to Experiment 1 with the exception that the sample and the response options were presented simultaneously for 2500 ms. Participants were able to respond at any point during the trial. 3.2. Results 3.2.1. Response times Error trials and trials in which participants failed to respond were excluded from the RT analyses, resulting in the elimination of 152 of the total 1920 trials (7.9% total). Fig. 4 shows the mean RT on correct trials plotted by Trial Type for Color and Word Targets. A two-way repeated measures ANOVA with Target Type (color targets, word targets) and Trial Type (baseline, congruent, incongruent-unrelated foil, and incongruent-related foil) as factors revealed a main effect of Target Type, F(1, 19) = 4.38, p = 0.05, ␩2 p = 0.19, and a main effect of Trial Type, F(3, 57) = 32.25, p < 0.001, ␩2 p = 0.63. The Target Type x Trial Type interaction was also significant, F(3,57) = 13.70, p < 0.001, ␩2 p = 0.42. To isolate the source of the Target Type x Trial Type interaction, separate one-way repeated measures ANOVAs with Trial Type as a factor were conducted for each Target Type. For Color Targets, there was a main effect of Trial Type, F(3, 57) = 35.12, p < 0.001, ␩2 p = 0.65. LSD post hoc tests on the Trial Type factor revealed that response times for Baseline trials were significantly lower than Congruent trials [t(19) = 5.67, p < 0.001, d = 1.26], Incongruent-Unrelated Foil Trials [t(19) = 6.87, p < 0.001, d = 1.53], and Incongruent-Related Foil Trials [t(19) = 6.730, p < 0.001, d = 1.50]. Response times for Congruent trials were significantly lower than Incongruent-Unrelated Foil [t(19) = 5.17, p < 0.001, d = 1.15], as well as Incongruent-Related Foil trials [t(19) = 5.13, p < 0.001, d = 1.14]. Incongruent-Unrelated Foil and Incongruent-Related Foil trials did not significantly differ from each other, t(19) = 0.07, p = 0.93, d = 0.01. For Word Targets, there was also a main effect of Trial Type, F(3, 57) = 3.74, p = 0.016, ␩2 p = 0.16. LSD post hoc tests on the Trial Type factor revealed that response times for Baseline trials were significantly lower than Congruent trials [t(19) = 2.93, p = 0.009, d = 0.65], Incongruent-Unrelated Foil Trials [t(19) = 2.69, p = 0.014, d = 0.60], and Incongruent-Related Foil Trials

60

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

Fig. 4. Mean response time on correct trials during Testing (in milliseconds) plotted by Trial Type for Color Targets (filled bars) and Word Targets (unfilled bars) for Experiment 2. Error bars represent standard errors of the means.

[t(19) = 2.67, p = 0.015, d = 0.59]. In contrast with the above analyses of Color Targets trials, response times for Congruent trials did not significantly differ from Incongruent-Unrelated Foil [t(19) = 0.71, p = 0.48, d = 0.16] or Incongruent-Related Foil trials [t(19) = 0.95 p = 0.35, d = 0.21]. Response times for Incongruent-Unrelated Foil and Incongruent-Related Foil trials did not significantly differ, t(19) = 0.04, p = 0.96, d = 0.01. 3.2.2. Error rates Trials in which participants failed to respond (7 of the total 1920 trials, 0.4% total) were excluded. Fig. 5 shows the mean error rate plotted by Trial Type for Color and Word Targets. A two-way repeated measures ANOVA with Target Type (color targets, word targets) and Trial Type (baseline, congruent, incongruent-unrelated foil, and incongruent-related foil) as factors revealed main effects of Target Type, F(1, 19) = 6.22, p = 0.02, ␩2 p = 0.25, and Trial Type, F(3, 57) = 22.60, p < 0.001, ␩2 p = 0.54. The Target Type x Trial Type interaction was significant, F(3, 57) = 21.13, p < 0.001, ␩2 p = 0.53. To isolate the source of the Target Type x Trial Type interaction, separate one-way repeated measures ANOVAs with Trial Type as a factor were conducted for each Target Type. For Color Targets, there was a main effect of Trial Type, F(3, 57) = 29.33, p < 0.001, ␩2 p = 0.61. LSD post hoc tests revealed that error rates for Baseline trials did not significantly differ from Congruent trials [t(19) = 0.29, p = 0.77, d = 0.06] or Incongruent-Unrelated Foil trials [t(19) = 1.02, p = 0.32, d = 0.22]. Baseline trials had significantly lower error rates than Incongruent-Related Foil trials, t(19) = 5.75, p < 0.001, d = 1.28. Error rates for Congruent trials did not significantly differ from Incongruent-Unrelated Foil trials, t(19) = 1.00, p = 0.30, d = 0.23. Error rates for Congruent trials were significantly lower than Incongruent-Related Foil trials, t(19) = 5.78, p < 0.001, d = 1.29. Error rates for IncongruentUnrelated Foil trials were significantly lower than Incongruent-Related Foil trials, t(19) = 5.35, p < 0.001, d = 1.19. For Word targets, there was no main effect of Trial Type, F(3, 57) = 1.17, p = 0.33, ␩2 p = 0.06. One-sample t-tests revealed that mean error rate was significantly less than chance (0.5) for all Trial Types and Target Types, ts(19) > 5.89, ps < 0.001. 3.3. Discussion The RT analyses indicated slowed responding when the sample was incongruent, regardless of the relatedness of the foil to the irrelevant dimension, but only on Color Target trials. This pattern of RT interference was equivalent for both incongruent trial types, independent of whether the foil was related or unrelated to the irrelevant sample dimension. On word target trials, no such interference was observed. Since the sample and response options were presented simultaneously, the relevant sample dimension was indicated by the response options dimension. This suggests that the irrelevant word response options undermined color-matching performance (Sabri et al., 2001). Interestingly, the results indicated interference on color-

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

61

Fig. 5. Mean error rates during Testing plotted by Trial Type for Color Targets (filled bars) and Word Targets (unfilled bars) for Experiment 2. Error bars represent standard errors of the means.

matching trials for both incongruent trial types regardless of foil-to-sample relatedness, which is similar to the pattern observed in Experiment 1 for color as well as word targets. The error rate data indicated a decrement in performance on color target trials only when the sample was incongruent and the foil response option was related to the irrelevant dimension of the sample. This result is similar to Experiment 1, except that it is only present on color matching trials. We interpret this as indicating that attention was biased towards a reading response in the simultaneous presence of sample and word response options. 4. General discussion In Experiment 1, participants were presented a bi-dimensional Stroop stimulus, the sample was removed, and immediately followed by the onset of target and foil response options. On average, participants were slower to respond to both color and word targets on trials when the Stroop stimulus was incongruent (e.g., the word “red” in blue font). In addition, error rates increased only on Incongruent – Related Foil trials. These results suggest that encoding and irrelevant affected responding, even when the foil response option did not match either dimension (Sabri et al., 2001). Additionally, the lack of increase in RTs for incongruent-related foil trials compared to incongruent-unrelated foils fails to support hypothesized response competition effects. This pattern is consistent with Sturz et al. (2013), which also found evidence for semantic but not response competition in the MTS task. In Experiment 2, participants were presented a bi-dimensional Stroop stimulus and the target and foil response options simultaneously. On average, participants were slower to respond to incongruent samples, but only for color targets. Critically, RT interference on incongruent trials was not significantly different regardless of the relatedness of the foil to the irrelevant sample dimension. No evidence of interference was observed on word-matching trials, suggesting that the simultaneous presentation procedure reintroduced automatic word-reading selective attention processes that the sequential procedure had eliminated. Similar to Experiment 1, error rates increased only on incongruent-related foil trials, though only for colormatching trials. Collectively, these patterns suggest that matching-to-sample with simultaneous presentation of sample and response options (i.e., no SOA) produced biased attention to the word dimension (Sabri et al., 2001). It is important to highlight that in Experiment 1 the relevant dimension was ambiguous prior to response option onset, but the relevant dimension was immediately cued at the start of each trial in Experiment 2, because the targets were visible from the outset of each trial. The differing patterns of results across the two tasks indicate that the psychophysical context of the task might contribute heavily to the interference effects on attention in this task. In short, the context in which both stimulus dimensions must be encoded and retained in memory (i.e., Experiment 1) resulted in a requirement of allocation

62

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

of attention to both dimensions, while the context in which immediate visual matching could be utilized to respond (i.e., Experiment 2) resulted in biased attention toward words over colors. Under conditions where the sample does not remain available, as in Experiment 1, the two sample dimensions must be retained, resulting in two active semantic codes, as well as the lexical and orthographical codes associated with both the word and the name of the color. Once the response options are presented, the irrelevant yet activated semantic code must be suppressed in order to select the target (Sturz et al., 2013; Zhang, Zhang, & Kornblum, 1999). The incompatible semantic codes should require suppression, which should slow responding, and the dual orthographic codes should increase the probability of selecting the foil, but only on trials where the foil was related to the irrelevant dimension of the sample. Alternatively, in Experiment 2, where the sample and response options were presented simultaneously, there was no need to retain both sample dimensions. Despite the psychophysical context of the simultaneous match-to-sample task, lexical, orthographical, and semantic information should still be processed prior to selection of a response option. Presumably, the immediate dimensional cuing available to participants in Experiment 2 produced an attentional bias for word reading. By systematically manipulating the psychophysical context (i.e., requiring or biasing attention to both or one of the sample dimensions) and the response context (i.e., through the relatedness of the foil response option to the irrelevant dimension), we provide evidence that the asymmetrical nature of Stroop-like interference in an MTS task might be modulated by topdown attentional control within this paradigm. When a word stimulus is presented, there is a very high probability that the word will be read, activating a stimulus-response relationship that has been very consistently established in the literature on Stroop and other Stroop-like tasks (Cohen et al., 1990; MacLeod, 1991; but see Besner et al., 1997). Given the extent to which the match-to-sample (MTS) Stroop-like task differs from more traditional Stroop tasks, it is important to highlight how our results diverge from past research. The most drastic divergence seems to be with respect to the source of the interference effect. Primarily, the source of interference in a traditional Stroop task appears to result from a combination of both semantic and response competition (De Houwer, 2003; Schmidt & Cheesman, 2005; Risko, Schmidt, & Cheesman, 2006). In contrast, the interference effect in the MTS task appears to be primarily a result of semantic competition (as evidenced here and in Sturz et al., 2013). We suggest that this difference may be the result of two components of the MTS task that are not involved in the traditional Stroop task. First, the delayed MTS task employs a sequential stimulus presentation. Second, regardless of delay, the relevant sample dimension in the MTS task remains ambiguous until attention is directed to the response options. The sequential stimulus presentation forces participants to attend to and retain both dimensions of the bi-dimensional sample until the relevant dimension is cued by response option onset. The sequential sample-then-response paradigm does not contain enough information for the participant to know which sample dimension to attend to and which to suppress; therefore, to perform the matching task, participants must select the relevant task dimension, color-matching or word-matching, at the moment of response option presentation. Such a paradigm is in stark contrast to the traditional Stroop task in which participants are explicitly instructed on the relevant sample dimension (for a review, see, MacLeod, 1991, 1992). We speculate that the MTS task component in which response options cued the relevant sample dimension may have been the critical factor in forcing attention to both sample dimensions. This forced attention to both sample dimensions (especially during a sequential presentation as in Experiment 1) presumably increased the role of semantic competition in the MTS task compared to a traditional Stroop task. By extension, such a task difference may have also been the reason for the lack of difference between Incongruent – Unrelated and Incongruent – Related trial types. Given that we have taken this lack of difference between these trial types as evidence against response competition in the MTS task, this may also explain why response competition is obtained in more traditional Stroop tasks but not in the current MTS task. It is worth noting that even when sample and response options were presented simultaneously (as in Experiment 2), and immediate dimensional cuing was available to participants (i.e., by virtue of the visibility of the targets at trial outset), we still obtained evidence consistent with semantic competition and inconsistent with response competition for the color matching condition. We speculate that even though immediate dimensional cueing was available to participants, they were still attending to both sample dimensions (given the top-down presentation of a trial structure). Although simultaneous presentation appears to have produced an attentional bias for word reading (and hence a more traditional Stroop-like effect), participants may still have been attending to both sample dimensions. We did not observe a typical facilitation effect, characterized as faster responding on congruent trials (e.g., the word “red” in red font) compared to baseline trials (Goldfarb & Henik, 2007). According to MacLeod and MacDonald (2000), the extent of Stroop facilitation is strongly related to the amount of practice with the task and integration versus separation of the stimulus dimensions. Though matching tasks have primarily been used to determine the role of response modality on the Stroop effect (Aarts, Roelofs, & van Turennouta, 2009; Gazzaley & Nobre, 2011; Steinhauser & Hübner, 2009), the results from the MTS tasks provide evidence that facilitation may be a function of top-down control for the task. Under the assumption that response competition occurs when two potential response codes compete for selection, we hypothesized that response competition should only be observed when the foil option was related to the irrelevant sample dimension (e.g., “red” in blue font with color response options blue and red). The irrelevant word “red” mapped to the color red (i.e., foil) and competed with the color blue (i.e., target) for response. The foil should not compete with the target on unrelated foil trials, because it did not map onto the retained codes (e.g., “red” in blue font with color response options blue and yellow). We found evidence for Stroop-like interference on both incongruent trial types, indicating competition regardless of whether the foil mapped to the relevant or irrelevant dimension. Assuming that requiring both sample dimensions to be retained resulted in encoding semantic codes for each dimension, we inferred that the pattern of results could only be explained

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

63

as semantic-based interference rather than response-based interference. As importantly, we attribute the decrement in accuracy on Incongruent-Related trials to the performance of the wrong task (i.e., match word to the name of the color) rather than evidence of a distractor response code competing for selection. We acknowledge that an alternative interpretation of our results may be that RTs reflect the presence of semantic competition whereas error rates reflect the presence of response competition. Specifically, the lack of difference in RTs between Incongruent – Related and Incongruent – Unrelated trials seems to be consistent with a semantic competition explanation whereas a difference between these trial types in error rates could be interpreted as consistent with a response competition account. Presumably though, any presence of response competition should have also manifested in RTs. As a result, we believe that a semantic competition explanation of RTs coupled with the possibility of increased error rates on Incongruent – Unrelated trials because of performance of the wrong task seems to be a feasible account of obtained results. Regardless, an interpretation of the presence of both semantic and response competition would be more consistent with explanations of the traditional Stoop effect. One caveat is that historical analyses of the Stroop effect are generally constrained to analyses of RTs, and it remains unclear whether such possibilities of one interpretation for RTs and another for error rates would emerge with the analysis of error rates under normal Stroop task conditions. In conclusion, we interpret our results as evidence that semantic-based informational conflict is largely responsible for the pattern of effects in a sequential match-to-sample Stroop task (i.e., Experiment 1; see also Sturz et al., 2013). The mechanism for overcoming the observed semantic conflict is likely inhibition, which refers to conflict of meaningfully similar colorrelated information that is incompatible within the matching-to-sample task. The sequential match-to-sample Stroop-like task likely produced semantic interference because no sample dimension was to be ignored at the input stage, but rather, the irrelevant sample dimension was activated, retained, and must be suppressed at the point that the relevant task dimension was identified. Although our conclusions about interference within the MTS task may not generalize to the canonical Stroop task, more typical patterns of Stroop interference, or theoretical explanations for the Stroop effect, it appears that results from the traditional Stroop task do appear to generalize to the MTS task, though only under specific conditions (i.e., simultaneous sample-response option onsets). Acknowledgements This research was conducted following the relevant ethical guidelines for human research and supported in part by funds from the Office of the Vice President for Research and the Jack N. Averitt College of Graduate Studies at Georgia Southern University to BRS and TWB. References Aarts, E., Roelofs, A., & van Turennout, M. (2009). Attentional control of task and response in lateral and medial cortex: Brain activity and reaction time distributions. Neuropsychologia, 47, 2089–2099. http://dx.doi.org/10.1016/j.neuropsychologia.2009.03.019 Augustinova, M., & Ferrand, L. (2012). Suggestion does not de-automatize word reading: Evidence from the semantically based Stroop task. Psychonomic Bulletin & Review, 19, 521–527. Augustinova, M., Flaudias, V., & Ferrand, L. (2010). Single-letter coloring and spatial cueing do not eliminate or reduce a semantic contribution to the Stroop effect. Psychonomic Bulletin & Review, 17, 827–833. Besner, D., Stolz, J. A., & Boutilier, C. (1997). The Stroop effect and the myth of automaticity. Psychonomic Bulletin & Review, 4, 221–225. Blais, C., & Besner, D. (2006). Reverse Stroop effects with untranslated responses. Journal of Experimental Psychology: Human Perception and Performance, 32, 1345–1353. Cohen, J. D., Dunbar, K., & McClelland, K. (1990). On the control of automatic processes: A parallel distributed processing account of the Stroop effect. Psychological Review, 97, 332–361. De Houwer, J. (2003). On the role of stimulus-response and stimulus–stimulus compatibility in the Stroop effect. Memory & Cognition, 31, 353–359. Gazzaley, A., & Nobre, A. C. (2011). Top-down modulation: Bridging selective attention and working memory. Trends in Cognitive Sciences, 16, 129–135. http://dx.doi.org/10.1016/j.tics.2011.11.014 Glaser, M. O., & Glaser, W. R. (1982). Time course analysis of the Stroop phenomenon. Journal of Experimental Psychology: Human Perception and Performance, 8, 875–894. Goldfarb, L., & Henik, A. (2007). Evidence for task conflict in the Stroop effect. Journal Of Experimental Psychology: Human Perception And Performance, 33(5), 1170–1176. http://dx.doi.org/10.1037/0096-1523.33.5.1170 Kahneman, D., & Henik, A. (1981). Perceptual organization and attention. In M. Kubovy, & J. R. Pomerantz (Eds.), Perceptual organization (pp. 181–211). Hillsdale, NJ: Erlbaum. Klein, G. S. (1964). Semantic power measured through the interference of words with color-naming. American Journal of Psychology, 77, 576–588. Luo, C. R. (1999). Semantic competition as the basis of Stroop interference: Evidence from color word matching tasks. Psychological Science, 10, 35–40. MacLeod, C. M., & Dunbar, K. (1988). Training and Stroop-like interference: Evidence for a continuum of automaticity. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14, 126–135. MacLeod, C. M., & MacDonald, P. A. (2000). Interdimensional interference in the Stroop effect: Uncovering the cognitive and neural anatomy of attention. Trends in Cognitive Sciences, 4, 383–391. MacLeod, C. M., Chiappe, D. L., & Fox, E. (2002). The crucial roles of stimulus matching and stimulus identity in negative priming. Psychonomic Bulletin & Review, 9, 521–528. MacLeod, C. M. (1991). Half a century of research on the Stroop effect: An integrative review. Psychological Bulletin, 109, 163–203. MacLeod, C. M. (1992). The Stroop task: The gold standard of attentional measures. Journal of Experimental Psychology: General, 121, 12–14. Posner, M. I., & Snyder, C. R. R. (1975). Attention and cognitive control. In R. L. Solso (Ed.), Information processing and cognition: The Loyola symposium (pp. 55–85). Hillsdale, NJ: Erlbaum. Risko, E. F., Schmidt, J. R., & Besner, D. (2006). Filling a gap in the semantic gradient: Color associates and response set effects in the Stroop task. Psychonomic Bulletin & Review, 13, 310–315. Sabri, M., Melara, R. D., & Algom, D. (2001). A confluence of contexts: Asymmetric versus global failures of selective attention to Stroop dimensions. Journal of Experimental Psychology: Human Perceptions and Performance, 27, 515–537. http://dx.doi.org/10.1037//0096-1523.27.3.515

64

M.L. Green et al. / Learning and Motivation 56 (2016) 53–64

Schmidt, J. R., & Cheesman, J. (2005). Dissociating stimulus–stimulus and response–response effects in the Stroop task. Canadian Journal of Experimental Psychology, 59, 132–138. Steinhauser, M., & Hübner, R. (2009). Distinguishing response conflict and task conflict in the Stroop task: Evidence from ex-Gaussian distribution analysis. Journal of Experimental Psychology: Human Perception and Performance, 35, 1398–1412. http://dx.doi.org/10.1037/a0016467 Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643–662. Sturz, B. R., Green, M. L., Locker, L., & Boyer, T. W. (2013). Stroop interference in a delayed match-to-sample task: Evidence for semantic competition. Frontiers in Psychology, 4, 1–10. http://dx.doi.org/10.3389/fpsyg.2013.00842 Zhang, H., Zhang, J., & Kornblum, S. (1999). A parallel distributed processing model of stimulus–stimulus and stimulus-response compatibility. Cognitive Psychology, 38, 386–432.