The Eriksen flanker effect revisited
Introduction
In a standard Eriksen flanker task (Eriksen & Eriksen, 1974) participants carry out a speeded response to a target letter which is closely surrounded by non-target letters or flankers. Choice reaction time (CRT) is invariably found to increase when flankers are incongruent – i.e. consist of letters from the target set that require another response – in comparison with conditions in which a target letter is presented alone or is surrounded by congruent flankers – i.e. letters from the target set that require the same response. When flankers are neutral – i.e. letters that do not belong to the target set – CRT may be slightly longer than in the congruent condition but the effect is much less pronounced than in the incongruent condition.
This pattern of results has been commonly viewed as support for information processing in terms of a continuous flow rather than in terms of transmission through a succession of discrete stages. (McClelland, 1979; Miller, 1988, Miller, 1993; Sanders, 1990, Sanders, 1998). Continuous flow entails considerable parallel activity when processing target and flankers, which, however, is no safeguard against interference due to, say, perceptual confusion of target and flankers or to competition among conflicting response tendencies (Eriksen & Eriksen, 1974; Eriksen & Schultz, 1979). Eriksen and co-workers interpreted their initial results in terms of conflicting stimulus evaluation of target and flankers, causing activation of their corresponding response channels, thus resulting in competition at the level of response activation. This view was supported by psychophysiological evidence. For instance, Coles, Gratton, Bashore, Eriksen, and Donchin (1985) found incongruent flankers to prolong EMG latency and to evoke EMG activity of the limb connected to the incorrect response. In addition, the lateralised readiness potential (LRP) appears to be initially biased towards the incorrect response side when flankers were incongruent (e.g. Ridderinkhof & van der Molen, 1995; Smid, Lamain, Hogeboom, Mulder, & Mulder, 1991). In addition, incongruent flankers appear to increase the P3 latency of the event-related brain potential (ERP), which has been widely viewed as evidence for increased demands on stimulus evaluation (Donchin & Coles, 1988; Coles et al., 1985; Ridderinkhof & van der Molen, 1995). Stoffels and van der Molen (1988) found additive effects of variation of visual congruence of target and flanker letters and of simultaneously presented irrelevant auditory location cues. This result was interpreted as evidence that the effect of visual congruity might be predominantly perceptual whereas irrelevant incongruent auditory location cues might frustrate response selection by triggering an inappropriate tendency to respond towards the location of the auditory stimulus. The effects of irrelevant auditory location cues and visual congruity appeared to interact when target and flankers consisted of arrows rather than of letters, which was interpreted as evidence for perceptual crosstalk between visual and auditory feature channels when both the visual and auditory stimuli contained spatial information.
Flowers and Wilcox (1984) and Flowers (1990) suggested perceptual facilitation in the case of congruent flankers. They varied stimulus onset asynchrony (SOA) and found incongruent flankers to interfere most at simultaneous presentation of target and flankers, which effect vanished when flankers preceeded the target by 200 ms. In contrast, congruent flankers hardly facilitated at simultaneous presentation but had a facilitating effect at an SOA of 200 ms. Flowers proposed that flankers might evoke two processes: (a) perceptual (repetition) priming which evolves as a function of SOA and facilitates when flankers are congruent, and (b) inhibition of a tendency to respond to flankers, which has the effect of eliminating interference at a longer SOA (Posner, 1978). In this way flankers would have both a perceptual and a response effect, although facilitation through repetition priming might be only effective in case of a SOA.
Perceptual confusion of target and flankers has been argued to depend on factors such as location information about and similarity among target and flankers (Estes, 1975; Bjork & Estes, 1973; Coles et al., 1985). Thus, Yeh and Eriksen (1984) studied the relative role of physical versus name identity on flanker interference. Among others, targets were surrounded by physical (e.g. AAA) or by name identical (e.g. QqQ) congruent flankers. In addition physical characteristics could be similar (e.g. eGe) or dissimilar (eAe). For example, congruent flankers could be name identical but physically dissimilar (e.g. aAa). Alternatively, flankers could be incongruent but physically similar (aGa), etc. The results showed a dominant role of physical features in regard to interference or facilitation. CRT was considerably faster when flankers and target were physically identical (AAA) than when they were name identical (aAa). Yet, CRT proved still somewhat faster when flankers were name identical than when they consisted of another congruent letter – i.e. a letter requiring the same response as the target – which suggests at least a contribution of name identity as well. Interference from incongruent flankers was less pronounced when target and flankers belonged to different physical letter categories (e.g. target: capital; flankers: small, or vice versa) than when they belonged to the same category (e.g. all capitals), suggesting that subjects discriminate targets and flankers primarily on the basis of their physical features. The relevance of physical characteristics was also clear from the pronounced interference that was observed when congruent or neutral flankers were physically similar to incongruent flankers. Yeh and Eriksen concluded, therefore, that responses are formost elicited by physical and not by name codes. This was taken as further evidence in favour of processing as a continuous flow rather than in discrete steps. Proponents of processing through discrete stages have commonly assumed that the final output of perceptual stimulus processing consists of a name code (e.g. Sanders, 1998). The Yeh and Eriksen result that response activation develops prior to achieving a name code is clearly at odds with this view. Yet, a name code as output of perceptual identification may be unwarranted. Recent data obtained using the additive factors method (AFM) suggest that the output of perceptual identification may be a holistic physical representation, whereas a name code is achieved during response selection (Sanders, van Duren, Wisman, Boers, & Klompenhouwer, submitted).
More generally, the evidence for perceptual confusion as contributor to flanker interference may not be fully watertight. For example, although Smid, Mulder, and Mulder (1990) confirmed that P3 latency was longer when a target was surrounded by flankers than when presented single, P3 latency proved to be insensitive to the similarity of flankers and target. Again, Smid, Mulder, Mulder, and Brands (1992) found no effect on P3 latency when stimuli were harder to distinguish (say, l (el) from 1(one)), which may cast some doubt on the P3 latency as an index of perceptual processing demands. As mentioned, an effect of repetition priming is probably limited to the case in which flankers and target are presented in succession. Other results suggesting perceptual confusion have relied on application of the AFM which is likely to be inappropriate in the case of composite stimuli as used in the Eriksen paradigm (e.g. Ridderinkhof, van der Molen, & Bashore, 1995; Sanders, 1998). In fact, most reviewed results might be accounted for by assuming full parallel and unrestrained perceptual analysis of target and flankers – i.e. no perceptual confusion – resulting in competing response tendencies.
A convincing test of a perceptual contribution to flanker interference requires a paradigm which enables separate assessment of processes involved in perceptual analysis and in response selection. This may be achieved by the paradigm of the functional visual field (e.g. Sanders, 1993, Sanders, 1998). In a prototypical study subjects inspect a stimulus presented at a fixed location in the left part of the visual field, (SL) followed by a shift of the eyes to a stimulus, presented at a symmetrical fixed location in the right part of the visual field (SR), both stimuli presented at eyelevel. SL and SR usually subtend a considerable binocular visual angle, say at least 45°. A trial ends with a joint choice reaction to SL and SR by way of a key press or a vocal response. In some studies the response has been a same–different response whereas in other studies SR indicated whether a response to SL should be carried out or be withheld. Measuring the start and the end of the saccade from SL to SR allows assessment of the fixation time of SL (TL), the time taken by the saccade (TM) and the time between the arrival of the eyes at SR and the completion of the response (TR).
There is accumulating evidence that perceptual variables, such as stimulus quality, affect CRT and TL about equally – suggesting that perceptual processing is tied to visual fixation. (Sanders & Houtmans, 1985; van Duren, 1994). In contrast, S-R compatibility has not been found to affect either TL or TR, suggesting that response selection may proceed in parallel with the saccade from SL to SR (van Duren & Sanders, 1995). It is tempting to assume that a saccade is triggered upon completion of perceptual stimulus analysis.
These and other results (see Sanders, 1998, pp. 196–214 for a review) suggest experiments on the effect of flanker congruity on TL. If the flanker effect is partly due to perceptual analysis, such as target localisation or target–flanker confusion, one would expect a flanker effect on TL. If flanker interference is fully a matter of response selection, flanker effects should be absent both in regard to TL and to TR.
Section snippets
Method and procedures
The experiment took place in a dimly illuminated, sound attenuated and electrically shielded cabin. Participants were seated at a table in front of a semicircular black screen which was enclosed by a black non-reflecting floor and ceiling. The screen was excised at eye level at three locations, i.e. at the centre, at 22.5° to the left and at 22.5° to the right of the visual meridian. A computer screen was positioned at all three excision areas. Participants held their head in a chin rest so as
Results and discussion
Incorrect reactions, errors of commission and times exceeding the mean by two standard deviations were excluded from analysis. Mean CRT (control condition) and mean TL, TM and TR (experimental condition) and error proportions are shown in Table 1
An ANOVA was carried out on the data of the control condition with individual mean CRT's as cells. The effect of flanker condition was highly significant , which confirms the commonly observed pronounced interference in the
Method and procedures
Testing took place in a sound attenuated room; stimuli were presented on a single Commodore 1936 screen, positioned at a distance of about 80 cm from the participants' eyes. Tining of events was controlled by the ERTS (version 3) system (Beringer, 1988). A trial started with presentation of the same 100 ms auditory warning tone and circular fixation light (during 1.5 s) as in Experiment 1. This was followed by two successively presented stimuli (S1 and S2). S1 was presented for 150 ms and
Results and discussion
The statistical analysis was based on correct responses only. The cells in the ANOVA were the individual means over sessions for flanker conditions, ISI durations and same vs different trials. The mean CRTs and error proportions as a function of flanker condition and ISI duration are presented in Table 2.
Flanker condition and ISI had a significant effect on CRT, whereas their interaction just failed to reach significance .
Method and procedures
The experimental setting and equipment were similar to those in Experiment 1. The control condition was even fully identical. The experimental condition differed from Experiment 1 in that, instead of a go/no go indication, SR consisted of a single letter (A or B). As in Experiment 2, the instruction was to carry out a same–different response in regard to the target of SL and the stimulus at SR, a same response corresponding to a left response key and a different response to a right response key.
Results and discussion
Mean CRTs of the correct reactions and error proportions, both averaged over participants, are shown in Table 4. Mean CRT for the same and different responses are presented in Table 5.
An ANOVA on CRT in the control condition with individual means as cells showed a significant effect of flanker condition . Error proportions differed significantly as well . Similar analyses on the measures TL, TM and TR of the experimental conditions failed to show any
Experiment 4
It is obvious that the absence of an effect of incongruent flankers on TL does not exclude conditions in which perceptual processing adds to interference. For instance, perceptual confusion of targets and flankers might only arise when they are physically similar (Yeh & Eriksen, 1984). Thus, TL may be affected when target and flankers are similar and not when they are dissimilar. Physical similarity might hamper target discrimination the processes of which are usually related to perceptual
Results and discussion
Mean TL, TR and error proportions are presented in Table 7. The mean TM values varied between 143 and 146 ms. A MANOVA on TL with the individual means as cells did not show any significant main effect or interaction of the variables. A similar analysis on the TR data showed a highly significant effect of the similarity of SL target and SR stimulus on the same–different CRT. Similar different responses (e.g. JJPJJ – F) took longer than dissimilar different responses (e.g.
General discussion
Taken together, the results showed the usual effect of incongruent flankers on CRT in the control conditions but neither on TL nor on TR in the setting of the functional visual field. The absence of an effect on TL, even not when target and flankers were physically similar, argues against a contribution to flanker interference on the levels of perceptual feature extraction and physical stimulus encoding. In contrast the results are consistent with the hypothesis that visual stimuli are
References (37)
- et al.
On the output of encoding during stimulus fixation
Acta Psychologica
(1988) Discrete and continuous models of human information processing: In search of partial output
Acta Psychologica
(1988)- et al.
On the robustness of the additive factors stage structure in blocked and mixed choice reaction designs
Acta Psychologica
(1988) - Beringer, J. (1988). ERTS Experimentelles Run Time System. Bericht des psychologischen Instituts der technischen...
- et al.
Letter identification in relation to linguistic context and masking conditions
Memory and Cognition
(1973) - et al.
Eye movements and stages of processing
Acta Psychologica
(1988) - et al.
Late positive components and stimulus evaluation time
Biological Psychology
(1981) - et al.
A psychophysiological investigation of the continuous flow model of human information processing
Journal of Experimental Psychology: Human Perception and Performance
(1985) - et al.
Is the P300 component a manifestation of context updating ?
The Behavioral and Brain Sciences
(1988) - et al.
Effects of noise letters upon the identification of a target letter in a non-search task
Perception and Psychophysics
(1974)
Response competition effects in same–different judgements
Perception and Psychophysics
Information processing in visual search: A continuous flow conception and experimental results
Perception and Psychophysics
The locus of interferential and perceptual processes
Journal of Experimental Psychology: General
Priming effects in perceptual classification
Perception and Psychophysics
The effect of flanking context on visual classification: The joint contribution of interactions at diferent processing levels
Perception and Psychophysics
Reducing the effect of adjacent distractors by narrowing attention
Journal of Experimental Psychology: Human Perception and Performance
On the time relations of mental processes: An examination of systems of processes in cascade
Psychological Review
The flanker compatibility effect as a function of visual angle, attentional focus, visual transients and perceptual load: A search for boundary conditions
Perception and Psychophysics
Cited by (83)
Effects of an acute bout of cycling on different domains of cognitive function
2024, Progress in Brain ResearchIdentifying cognitive deficits in cocaine dependence using standard tests and machine learning
2019, Progress in Neuro-Psychopharmacology and Biological PsychiatryMore abstract, more difficult to shift: Behavior and electrophysiological evidence
2019, Behavioural Brain ResearchBrain responses associated with different hierarchical effects on cues and targets during rule shifting
2018, Biological PsychologyCitation Excerpt :N200 and N300 were defined in term of the peak latencies. Previous studies have proposed that the frontal areas are responsible for task switching (Barceló, Periáñez, & Knight, 2002; Groom & Cragg, 2015; Sanders & Lamers, 2002; Vogel & Luck, 2000); thus, the frontal electrodes were first chosen for analysis. In order to study the dynamic response of the whole brain, we added the parietal electrode sites.