Elsevier

Brain Research

Volume 1179, 7 November 2007, Pages 106-118
Brain Research

Research Report
Cueing method biases in visual detection studies

https://doi.org/10.1016/j.brainres.2007.08.032Get rights and content

Abstract

Questions about attention are usually addressed by cueing tasks assessing whether knowledge of stimulus-related information provided in advance will improve target processing. Here, we test the reliability of this classical paradigm by means of using neutral cues in a simple visual detection task. We compared “mixed-block” (cued/no-cued trials are intermixed in the same block of trials) to “pure-block” (cued/no-cued trials are presented separately) protocols. We report converging evidence with behavioral and fMRI experiments that cueing methods entail competing processes of automatic motor activation (triggered by the cue) and proactive response inhibition (intended to counteract automatic responses to the cue). This competition strongly affects the reaction time baseline necessary to measure the “cueing” effect in a mixed-block design. Indeed, in such a protocol, proactive inhibition cannot be released before target presentation. Accordingly, we suggest that this design inevitably leads to biases in interpreting cueing data, and that the effects classically observed with mixed-block protocols are likely to be artifacts that are not attentional in origin. We conclude that the identification of this methodological issue now calls for a reassessment of the theoretical framework used to interpret some cueing effects with respect to their control baseline.

Introduction

Attention is supposed to modulate and improve sensory brain activity before the onset of an incoming stimulus when the observer can prepare for it (‘baseline shifts’, Driver and Frith, 2000). In recent years, advances in cognitive psychology and neuroscience have identified three attentional networks that involve different types of attention using a derived cueing method (Posner and Petersen, 1990): orienting, executive control and alerting. Orienting refers to the selection of information from sensory input. Executive control is defined as resolving conflict among responses. Finally, alerting carries out a more general function, involving a change in the internal state in order to become ready for any incoming task-related event. Posner's methods and model have long been the most influential in the field (see Posner, 2004 for an illustration of this influence on the whole cognitive neuroscience of attention). According to this model, three distinct and specific neural networks may be defined both in anatomical and functional terms (Fan et al., 2005): alertness would rely on a relatively large fronto-temporo-parietal network (inferior frontal gyrus, superior temporal and fusiform gyrii, inferior and superior parietal lobes) along with the thalamus. Orienting would also rely on a distributed fronto-temporo-parietal network. However, it would be relatively separated and functionally independent from the alerting network despite it shows some overlap (superior frontal and precentral gyrii, fusiform gyrus, superior parietal lobe and postcentral gyrus). Finally, the executive network would mainly involve the anterior cingulate cortex and frontal areas (superior, middle and inferior frontal gyrii), but would also activate the fusiform gyrus and the thalamus.

How are these effects of visual attention measured? This question is usually addressed by cueing tasks assessing whether knowledge of stimulus- or response-related information provided in advance will improve target processing. It is generally estimated by reaction time – RT – tasks. Simple RT is widely used as an efficient tool to study the nature of the mental processes and structures involved in visual detection. It requires detecting and providing a rapid and unvarying response to a constant target that occurs over a prolonged period of time at a relatively infrequent rate (Stuss et al., 2005). While choice RT clearly requires substantial monitoring and inhibition to prevent wrong responses (i.e., is particularly adapted to study executive control: Burle et al., 2002, Burle et al., 2004, Stuss et al., 2002, Stuss and Anderson, 2004), simple RT is not supposed to do. It is therefore more adapted to study the effects of orienting of visual attention, alerting or any other aspect of the energization process it depends on (initiation and constant activation of the strength of the response; Stuss and Alexander, 2000; see also Sanders, 1998).

A typical RT paradigm usually includes a cue before the response stimulus is presented. Depending on the function which is under scrutiny, the cue can be spatially informative (orienting, spatial cue), can introduce conflicts with respect to target identification, decisional or programming processes (executive control, e.g., incongruence between cue and target identifications) or can just give advance information about the temporal window of the forthcoming target (alerting, neutral cue). After a variable cue–target delay (foreperiod), the target is presented. In the classical Posner's method, the behavioral analysis of the efficiencies of the three attentional networks is based on the comparison of the performance of each cueing condition to its own reference condition (Attentional Network Test, ANT, Fan et al., 2002, Fan et al., 2005). The orienting effect is measured by subtracting spatially cued RT to neutrally cued RT. The conflict effect is estimated by contrasting incongruent with congruent conditions (the target – an arrow pointing leftward or rightward – is “flanked” by other arrows pointing either in the opposite or in the same direction, respectively). Finally, the alerting effect is measured by means of a comparison of neutrally cued RT to a particular condition in which no cue is presented. In the ANT, all these conditions are intermixed in the same block of trials. This test is widely used in Psychological research because it is robust and provides a behavioral measure of the efficiency of the three attentional networks within a single task and with the minimum number of trials. In addition, since each condition can be paired with its own control, it is convenient to take advantage of the unique design of the ANT to adapt it to neuroimaging (e.g., Fan et al., 2005).

Both for behavioral, neurophysiological and anatomical models, most of our knowledge of the functional architecture of attention was built on such cueing protocols (see Posner, 2004 for illustration). However, two main issues arise from the global cueing method. First, some protocols do not use any baseline to control for the cueing effect. Second, when present, the baseline used to compute the cueing effect may have a variable meaning across studies. These issues are likely to introduce biases in understanding and interpreting cueing data in visual detection studies. Accordingly, despite its apparent simplicity, what takes place during the foreperiod of simple RT (i.e., when subjects must withhold responses while waiting to detect the trigger stimulus) still remains relatively unclear. In this paper, we suggest that cueing methods may be partly responsible for some old theoretical confusion and controversies.

Since Niemi and Näätänen (1981), it is accepted that a warning signal (in other terms, a neutral cue) and its concomitant variable foreperiod may have multiple and interrelated effects on simple RT. However, in this review still widely cited, the authors suggest that the subject's expectancy (momentary probability of the immediate delivery of the response signal) is the most important determinant of the preparation to respond to the target at any moment during a trial and, hence, of RT. Accordingly, many studies were interested only in the relation of foreperiod duration to RT (e.g., Bertelson, 1967, Ollman and Billington, 1972, Miniussi et al., 1999). In other words, many studies have focused on the effect of presenting a warning signal only by considering the evolution of RT according to the cue–target delay and did not use any baseline to control for the cueing effect. Most interpretations of the foreperiod duration/RT relation (the delay dependent speeding) suggest a facilitation effect of the warning signal. Some assume it is mainly due to a process of temporal anticipation based on the increasing probability that the stimulus will occur during the foreperiod (see Niemi and Näätänen, 1981 for review, but also Ratcliff et al., 1999). Others have more recently interpreted the FP–RT function as arising from sequential effects of temporal variation of preparatory FP (Los and Heslenfeld, 2005, Van der Lubbe et al., 2004: responses would be slower when the foreperiod is shorter on the current than on the previous trial. Accordingly, differential sequential influences on mean RT would be observed at early and late FP – short FP would generate longer RT – and the FP–RT function could be interpreted as trace conditioning effects).

However, since Posner's influent methodological considerations (e.g., Posner, 1980), it is admitted that no-cued trials need to be considered as a reference to interpret the temporal aspects of the preparatory period in terms of a balance between RT costs and benefits. In this classical paradigm using exogenous cues, both orienting and alerting benefits are observed (e.g., Fan et al., 2002, Fan et al., 2005, Fernandez-Duque and Posner, 1997, Posner, 1980, Posner and Petersen, 1990). However, some conflicting results were also reported. For instance, some studies failed to provide direct evidence for cue RT facilitation (e.g., Tassinari et al., 1994). Others suggest that the cue might have a negative effect on RT with respect to the no-cued condition. In simple detection tasks, a response to a target presented very soon after the cue may be slowed down (Van Der Lubbe et al., 1996, Van Der Lubbe et al., 2005, Van Der Lubbe et al., 2006). Dealing specifically with spatial orienting, the authors have provided evidence that, for short foreperiods, cueing effects would rather be due to a speeded motor inhibition triggered by the spatial visual cue than to the withdrawal of attention as usually interpreted. It is likely that this effect also applies when using neutral cues. RT increase after cue presentation may also occur because the response of the visual system to the target is reduced by the previous stimulation from the cue (reduced sensitivity). This is a bottom-up, sensory effect, independent of orienting (Tassinari and Berlucchi, 1993; see also Berlucchi, 2006 for review). In other words, the interpretation of exogenous cueing effects may not necessarily or uniquely concern the influence of attention on perception, whatever the cue is spatially informative or not.

Whereas the delay dependent speeding of RT is a highly reproducible phenomenon, the interpretation of the effect of a neutral warning signal strongly depends on task and experimental conditions. For example, it was pointed out recently that some controversial fMRI results might originate from the various behavioral paradigms which are used with regard to how cued and no-cued trials are mixed in the experimental blocks (Thiel et al., 2004).1 In the abovementioned paper, the criticism applied to alertness studies only. Despite the fact that potential baseline biases especially concern studies using warning signals (neutral cues), all other aspects of attention can potentially be affected by the way cues are presented. This methodological consideration is obviously critical in imaging studies since baseline shifts or sensory responses induced by the cue and stimulus-locked modulations of brain activity may be confounded as previously suggested (Driver and Frith, 2000, Liu et al., 2005). But it is also highly critical at a behavioral level. Indeed, if the way cued and no-cued trials are mixed in the experimental blocks affects RT, then it is likely that the theoretical interpretation of cueing effects may be biased.

Nonetheless, the problem of pure-blocks vs. mixed-blocks of trials in experimental designs is not new in cognitive psychology (e.g., Posner and Cohen, 1984). Faster RTs are generally observed when the levels of the independent variable are presented isolated from each other in pure-blocks than when they are presented randomly intermixed in mixed-blocks. The fact that, broadly, many additional factors are possibly confounded in this “mixing cost” effect was suggested by Los (1996, see for review): a strategic view holds that subjects are less well prepared in mixed-blocks than in pure-blocks, due to greater uncertainty about the condition to be presented on the forthcoming trial. A stimulus-driven view holds that residual activation stemming from a preceding trial has less influence in mixed-blocks than in pure-blocks due to greater intertrial variability. An energetic view considers that mixed- and pure-blocks might also differ on dimensions like effort and arousal. Finally, the mental system might be more “loaded” in mixed-blocks than in pure-blocks, due to the mere requirement to maintain readiness of all mental structures that could be called upon by either level of the independent variable.

However, these interpretations remain at the level of task difficulty and attentional resources. What specific processes could be responsible for this mixing cost remains unclear. Nevertheless, it is likely that both activation and inhibitory mechanisms are involved in the performance of cued trials in visual detection studies (Van Der Lubbe et al., 1996, Van Der Lubbe et al., 2005, Van Der Lubbe et al., 2006). We suggest that the interaction of these competitive effects could influence the performance of no-cued trials according to how cued and no-cued trials are mixed in the experimental design. Such a potential bias has not clearly been demonstrated and the mixed-block design is still mostly used in visual detection studies. This is a major methodological issue on which rely possible biases in understanding and interpreting cueing data.

In summary, there is a consensus in the literature about the alertness function and its behavioral correlates. A general and unfocused state of alertness is usually provided by a non-spatially informative cue given before the target. Such an alerting cue is supposed to speed up target detection by means of a brief automatic surge in arousal (state of physiological reactivity) mediated by the norepinephrine network (Coull, 1998, Coull et al., 2001, Witte and Marrocco, 1997). In the following series of experiments, we use this simple effect to test the reliability of the classical exogenous cueing paradigm with regard to how trials are mixed in the experimental blocks. Indeed, the alerting mechanism is supposed to be very consistent and reliable, as illustrated by its spatial extent or by the variety of sensory inputs which are able to trigger the effect (e.g., Fernandez-Duque and Posner, 1997).

In Experiment 1a, we provide subjects with non-spatially informative cues and measure the extent of the expected “alerting” effect. Such a warning signal is supposed to reduce RT for target detection by comparison with a control condition in which no cue is given prior to target presentation. However, the neutral cue and the control conditions are usually mixed in the same block of trials. In other words, this classical Posner-like paradigm is based on the assumption that this difference in RT solely reflects a change in the alert state elicited by the presentation of the neutral cue. We challenge this view and propose that the so-called ‘alerting effect’ might be a bias generated by the unpredictability of cue occurrence in the standard mixed-block protocol. More precisely, we hypothesize that inhibition is required to counteract potential responses to the cue when the protocol includes uncertainty about the nature of the forthcoming stimulus (cue or target). We predict that this may strongly affect RT with regard to pure-block designs, especially for the no-cued condition (i.e., when the target is presented unexpectedly without being preceded by a cue).

In order to provide further behavioral evidence of the involvement of inhibitory mechanisms in the mixed-block cueing design, Experiment 1a was also adapted in order to fit a classical Go/NoGo task in a complementary experiment (1b). It is well known that nogo trials induce inhibition but that the insertion of nogo signals also delays response latency on go trials. Theoretically, withholding a response to the nogo stimulus may not necessarily require an active inhibitory process (e.g., De Jong et al., 1995, Van den Wildenberg et al., 2002, Logan, 1994). However, as pointed out recently by Picton and colleagues (2007), the inhibitory processes involved in the go-nogo paradigm when any stimulus may both initiate or stop the response are similar to those involved in the stop-signal paradigm (e.g., Aron et al., 2003). In other words, the setup used in Experiment 1b aims at revealing an active inhibitory process rather than a decrease in response readiness.

Finally, Experiment 2 illustrates the origin of the bias with event-related fMRI by (i) evidencing cue-induced automatic motor activations (i.e., the need to implement proactive inhibition), and (ii) identifying the structures involved in the baseline effect.

Section snippets

Errors

For each subject, the total amount of errors was less than 5% for each block, as requested by the experimental constraints. In this analysis, we have collapsed false alarms and anticipations (abnormally short RT, see methods section). Results (mean number of errors) are presented in Fig. 1. Since there was no variance in some conditions, we could not apply conventional statistical tests and rather used t-tests when necessary. In the mixed-block condition, all of the errors were induced by

Experiment 1

In the mixed-block design, the classical RT advantage of neutrally cued with respect to no-cued trials was observed. This advantage is usually attributed to alertness benefits. In the pure-block design, however, we found no benefit of visual alerting on visual target detection, but rather an increase in RTs for short cue–target delays. In other words, cueing would generate cost rather than facilitation when interpreting the results only on the basis of pure-blocks data (Fig. 2). Interestingly,

Participants

Twelve right-handed males (aged 21 to 36 years), with normal or corrected to normal vision and without history of neurological or psychiatric disease, gave informed consent to participate in the experiment.

Task design and procedure

In this experiment, the paradigm consisted of a cued target detection task adapted from Posner's general method (Fig. 6). We have tested the alerting effect by using peripheral neutral cues in two ways. First, cued and no-cued trials were mixed up randomly in the same block of trials

Acknowledgments

The authors are grateful to Giovani Berlucchi and Carlo-Alberto Marzi (Dipartimento di Scienze Neurologishe e della Visione, Verona, Italia) for their help in conceiving the project and discussing data. They are also grateful to Jean-Luc Anton, Muriel Roth and Bruno Nazarian (fMRI center, CNRS Marseille) for invaluable help and direct contribution to acquisition and analysis of Experiment 2 data. This work was funded by the French Ministry of Research (A.C.I. JC 6042 to PB). Magali Jaffard was

References (62)

  • K. Konrad et al.

    Development of attentional networks: an fMRI study with children and adults

    Neuroimage

    (2005)
  • T. Liu et al.

    Transient attention enhances perceptual performance and FMRI response in human visual cortex

    Neuron

    (2005)
  • S.A. Los

    On the origin of mixing costs: exploring information processing in pure and mixed blocks of trials

    Acta Psychol.

    (1996)
  • J.W. Mink

    The basal ganglia: focused selection and inhibition of competing motor programs

    Prog. Neurobiol.

    (1996)
  • R. Näätänen et al.

    Timing of expectance peak in simple reaction time situation

    Acta Psychol. (Amst)

    (1974)
  • N.S. Narayanan et al.

    Reversible inactivations of rat medial prefrontal cortex impair the ability to wait for a stimulus

    Neuroscience

    (2006)
  • R.T. Ollman et al.

    The deadline model for simple reaction times

    Cogn. Psychol.

    (1972)
  • J. Requin et al.

    Neural correlates of partial transmission of sensorimotor information in the cerebral cortex

    Acta Psychol.

    (1995)
  • K. Sasaki et al.

    Suppression of visually initiated hand movement by stimulation of the prefrontal cortex in the monkey

    Brain Res.

    (1989)
  • D.T. Stuss et al.

    The frontal lobes and theory of mind: developmental concepts from adult focal lesion research

    Brain Cogn.

    (2004)
  • D.T. Stuss et al.

    Multiple frontal systems controlling response speed

    Neuropsychologia

    (2005)
  • G. Tassinari et al.

    Sensory and attentional components of slowing of manual reaction time to non-fixated visual targets by ipsilateral primes

    Vis. Res.

    (1993)
  • G. Tassinari et al.

    Do peripheral non-informative cues induce early facilitation of target detection?

    Vis. Res.

    (1994)
  • C.M. Thiel et al.

    Cerebral correlates of alerting, orienting and reorienting of visuospatial attention: an event-related fMRI study

    Neuroimage

    (2004)
  • W.P. Van den Wildenberg et al.

    Reduced response readiness delays stop signal inhibition

    Acta Psychol. (Amst)

    (2002)
  • R.H. Van Der Lubbe et al.

    Threefold effect of peripheral precues: alertness, orienting, and response tendencies

    Acta Psychol.

    (1996)
  • R.H. Van der Lubbe et al.

    Being prepared on time: on the importance of the previous foreperiod to current preparation, as reflected in speed, force and preparation-related brain potentials

    Acta Psychol.

    (2004)
  • A.R. Aron et al.

    Stop signal inhibition disrupted by damage to right inferior frontal gyrus in humans

    Nat. Neurosci.

    (2003)
  • G. Berlucchi

    Inhibition of return: a phenomenon in search of a mechanism and a better name

    Cogn. Neuropsychol.

    (2006)
  • J.R. Booth et al.

    Larger deficits in brain networks for response inhibition than for visual selective attention in attention deficit hyperactivity disorder (ADHD)

    J. Child Psychol. Psychiatry

    (2005)
  • M. Brett et al.

    Region of interest analysis using an SPM toolbox

  • Cited by (61)

    • Prefrontal contributions to action control in rodents

      2021, International Review of Neurobiology
    • Practice-induced functional plasticity in inhibitory control interacts with aging

      2019, Brain and Cognition
      Citation Excerpt :

      The activity at 150 ms thus entails preparatory processes engaged as soon as a task-relevant stimulus was discriminated, before the decision to actually engage inhibition processes (Criaud & Boulinguez, 2013). Together with the behavioral results suggesting more cautious response strategy in older adults, the timing, location and direction of this modulation suggests that with practice the young tended to rely less on proactive inhibitory control, most likely because of an improvement in their capacity for reactive inhibition (Jaffard, Benraiss, Longcamp, Velay, & Boulinguez, 2007; Jaffard et al., 2008). In contrast, the older adults showed no change in activity in this area.

    View all citing articles on Scopus
    View full text