Transcranial direct current stimulation (tDCS) reveals a dissociation between SNARC and MARC effects: Implication for the polarity correspondence account

Abstract

The concept of stimulus response compatibility (SRC) refers to the existence of a privileged association between a specific stimulus feature and a specific response feature. Two examples of SRC are the Spatial Numerical Association of Response Codes (SNARC) and the Markedness Association of Response Codes (MARC) effects. According to the polarity correspondence principle, these two SRC effects occur because of a match between the most salient dimensions of stimulus and response. Specifically, the SNARC effect would be caused by a match between right-sided responses and large numbers, while a match between right-sided responses and even numbers would give rise to the MARC effect. The aim of the present study was to test the validity of the polarity correspondence principle in explaining these two SRC effects. To this end, we applied transcranial direct current stimulation (tDCS) over left and right posterior parietal cortex (PPC), which is thought to be the neural basis of salience processing, during a parity judgement task. Results showed that cathodal tDCS over the PPC significantly reduced the MARC effect but did not affect the SNARC effect, suggesting a dissociation between the two effects. That is, the MARC would rely on a salience processing mechanism, whereas the SNARC would not. Despite this interpretation is in need of further experimental confirmations (i.e., testing different tasks or using different tDCS montages), our results suggest that the polarity correspondence principle can be a plausible explanation only for the MARC effect but not for the SNARC effect.

Introduction

The term stimulus–response compatibility (SRC) was coined more than 60 years ago (Fitts & Seeger, 1953) and refers to the phenomenon by which responses are faster and more accurate when stimuli are paired to responses that are physically or conceptually similar to, or have dimensional overlap with the stimuli (Kornblum, Hasbroucq, & Osman, 1990 in; Yamaguchi & Proctor, 2012).

More specifically, that of SRC is a concept that refers to the existence of a privileged association between a specific stimulus feature and a specific response feature. It has been explained by different theoretical frameworks, which mostly refer to the existence of a dual-route, one automatic and one non-automatic, or controlled. One of these models is, for example, that of Kornblum et al. (1990), who proposed that, when there is a similarity between the dimensions along which the stimulus and response sets vary, then a stimulus wil1 activate its corresponding response via an automatic route. However, only if this response is verified to be correct via the non-automatic route, i.e., the controlled one, it can be executed immediately and responding is facilitated. In contrast, if the outcome of the verification process signals that the response is incorrect, the response must be aborted and responding is inhibited (Proctor, Lu, Wang, & Dutta, 1995; see also; De Jong, Liang, & Lauber, 1994). This inhibition results, therefore, in a slower and less accurate response, respect to when response is facilitated (i.e., faster and more accurate) because of SRC.

SRC is a very robust effect: it can be observed with various types of stimulation and response modes, and can be induced also by task-irrelevant stimulus and response features (for a review see Proctor & Vu, 2006). Also, because it can be produced in a wide range of conditions, both experimental and ecological, like driving or piloting (Yamaguchi and Proctor, 2006, Yamaguchi and Proctor, 2012), SRC represents one of the major principles considered in designing computerized tools, which can effectively interact that with the human cognitive system (Proctor and Van Zandt, 2008, Proctor and Vu, 2006, Wickens et al., 2004).

A well-known example of SRC is the Simon effect (Simon & Small, 1969). In a typical Simon task participants are required to respond to the colour of the stimuli presented to the right or to the left on the computer screen, by pressing a left or a right key. Although stimulus location is irrelevant to the task, responses are faster and more accurate when stimulus and response locations correspond than when they do not correspond. The Simon effect is the result of the facilitation in responding to stimuli on the right side with the right-side response, and to stimuli on the left side with the left-side response than vice-versa (see Hommel, 1997, Hommel, 2011, Umiltá and Nicoletti, 1990).

Another example of SRC is the Spatial–Numerical Association of Response Code (SNARC) effect (Dehaene, Bossini, & Giraux, 1993), which depends on the preferential associations, signalled by faster and more accurate responses, between relatively small/large numbers and left/right responses.

The SNARC effect has been considered to be not a simple variant of the Simon effect (Mapelli, Rusconi, & Umiltà, 2003), and is actually thought to be the result of representing numbers spatially, i.e., along a mental number line, in which small numbers would be located on the left and large numbers on the right, at least in people with a left-right writing system (Dehaene et al., 1993; for a recent review see; Göbel, Shaki, & Fischer, 2011). Therefore, a preferential mapping would exist between right-hand responses and comparatively large numbers and between left-hand responses and small numbers (Landy, Jones, & Hummel, 2008).

However, SRC effects have been obtained also with set of stimuli, or stimulus dimensions, with no obvious spatial feature, as, for example, number parity. This is the case of the Markedness Association of Response Codes (MARC) effect (Willmes & Iversen, 1995; see also; Berch et al., 1999, Reynvoet and Brysbaert, 1999), i.e., the preferential association between the parity of a number and the response side. Precisely, responses are faster and more accurate when even/odd numbers are associated with right/left responses, with respect to the opposite pairing (Nuerk et al., 2004, Willmes and Iversen, 1995).

Despite the fact that it has received less research attention in comparison to the Simon effect and the SNARC effect, the MARC effect is a particularly interesting instance of SRC. That is because in the MARC effect the association does not concern only a spatial feature, i.e., that of the response side, but it originates from its preferential association with a dimension that, by itself, would modulate the response, i.e., number parity. Actually, much evidence has shown that the distinction between odd and even numbers is crucial, not only mathematically but also psychologically. Even digits are processed faster and/or more accurately than odd digits, and this preference emerges regardless of whether the digits are presented as Arabic numerals, rows of dots, or spoken words (Hines, 1990). Therefore, the MARC effect, along with several other compatibility effects (for a review see Alluisi & Warm, 1990), suggests that SRC is probably based on the association between stimulus and response features that are already considered as “preferred” by the cognitive system.

This is what suggests one of the most interesting accounts proposed to explain different forms of SRC: the principle of polarity correspondence (Proctor & Cho, 2006; for a much older hint at this principle, see; Chase and Clark, 1971, Seymour, 1973). According to the principle of polarity correspondence, SRC would result from a more general, not necessarily spatial, principle of stimulus–response mapping.

This principle posits that, in the binary representation of stimulus and response dimensions, the salient dimension plus polarity (+ polarity) is generally more available, and thus easier to process, than the non-salient dimension, minus polarity (− polarity).

In addition, when applied to the case of two bipolar dimensions, the polarity correspondence principle predicts that a processing advantage specifically arises in those conditions in which the two polar signs, i.e., two salient stimuli/features, match (Clark, 1969, Proctor and Cho, 2006).

It is important to point out that, in this context, saliency does not correspond to the relative relevance of a stimulus dimension, and, therefore, it is totally independent of the task employed and of the request of the task itself. Hence, according to the polarity correspondence principle, SRC effects would be the consequence of the correspondence between a salient response feature, and a salient response, and, therefore, between the non-salient (or less salient) stimulus and the non-salient (or less salient) response.

This account is perfectly compatible with the dual-process models. Those models are able to explain the reason why SRC effects arise from the automatic processing of preferential associations, and the reason why automatic processing is replaced by a controlled process in the case of non-preferential associations. In contrast, the polarity correspondence account can explain why certain associations exists. That is, SRC would be the consequence of pairing stimuli and responses that share the feature of being both salient (or of being both non-salient), rather than pairing a salient stimulus with a non-salient response or vice versa.

No doubt, the principle of polarity correspondence has a merit: it is very general and, because of that, it can explain the effects of a variety of associations between stimulus and response features, irrespective of whether the features involved are spatial or not. Based on this principle, even and large digits can be considered as more salient than odd and small digits and, similarly, the right side, of both stimulus and response, represents the preferential (the salient) location when compared with the left.

Simon, SNARC and MARC effects can be easily accounted for on the basis of polarity correspondence: a stimulus presented on the right side, a large number, or an even number are all preferentially associated with right responses because they are all cognitively “salient”. However, despite its parsimony, the explanatory power of the polarity correspondence principle has been questioned in the case of the SNARC effect.

In an interesting study by Leth-Steensen and Citta (2016), two groups of participants were asked to make even–odd parity judgements, using either “left” and “right” or “bad” and “good” vocal responses, and the presence of both the SNARC and the MARC effects was investigated.

According to the polarity correspondence principle, “right”, “good”, “large” and “even” would be more salient (+polarity), whereas “left”, “bad”, “small” and “odd” would be non-salient.

If the SNARC and MARC effects are attributable to polarity correspondence-driven SRC, they were expected to occur for both groups of participants, with faster left/bad responses for odd/small numbers, and faster right/good responses for even/large numbers, assuming the two response alternatives in this binary judgement task can be polarity coded.

However, contrary to expectation, in the group of participants instructed to give left/right responses, only the SNARC effect was present, while the MARC effect was absent. In contrast, in the bad/good response condition, while the MARC effect was robust, the SNARC effect was absent (Leth-Steensen & Citta, 2016).

This dissociation suggests that these two SRC effects, the SNARC and MARC effects, are based on two different mechanisms and that the polarity correspondence account would we valid only to explain the MARC effect, but not the SNARC effect (Leth-Steensen & Citta, 2016).

The Leth-Steenses and Citta's (2016) study is very interesting from a theoretical point of view, because it provides an important constraint on the presumed generality of the polarity correspondence account. However, since it is based on fragile results, as stated by the authors themselves, further evidence is needed in order to confirm the notion that the MARC and the SNARC effects dissociate.

With the present study we wanted to shed light on this recent debate, providing evidence in support or against the validity of the polarity correspondence principle in explaining both the SNARC and the MARC effects.

In order to do so, we investigated whether these two SRC effects are differentially influenced by the non-invasive stimulation of the posterior parietal cortex (PPC), a region involved, not only in attentional control mechanisms (Ashbridge et al., 1997, Corbetta and Shulman, 2002, Humphreys, Romani, Olson, Riddoch, & Duncan, 1994, Nobre et al., 2003, Wojciulik and Kanwisher, 1999), but also in salience-based attentional mechanisms (Bardi et al., 2013, Hodsoll et al., 2009, Kanai et al., 2011, Mevorach et al., 2010, Mevorach et al., 2006, Mevorach et al., 2009).

In the study of Bardi et al. (2013), transcranial direct current stimulation (tDCS), with an intensity of 1.5 mA, was applied bilaterally over the posterior parietal cortex (PPC). Enhancement of brain activity in the right PPC, via online anodal tDCS (and concurrent cathodal stimulation of the left PPC) improved attention to salient stimuli.

As tDCS has been successfully employed to modulate salience processing (Bardi et al., 2013), we tested the effect of tDCS over the PPC on the SNARC and the MARC effects elicited during a parity judgement task. With tDCS, a weak electrical current is applied to the scalp, inducing a modulation of the resting membrane potentials of neurons in the underlying cortex. Anodal stimulation causes membrane depolarization and enhances cerebral excitability, whereas cathodal stimulation diminishes it (Jang et al., 2009, Nitsche and Paulus, 2000, Nitsche and Paulus, 2001, Stagg et al., 2009). The advantage of using brain stimulation is that, unlike imaging techniques, which are purely correlational in nature, brain stimulation can inform us about a causal relation between the stimulated brain area and a possibly related cognitive function.

In detail, because both left and right PPC appear to be both involved in salience processing but with different roles, i.e., the right PPC is supposed to be involved in the processing of the more salient stimulus feature (+polarity) whereas the left PPC would be involved in the processing of the less salient feature (−polarity) (see Bardi et al., 2013, Hodsoll et al., 2009, Kanai et al., 2011, Mevorach et al., 2009, Mevorach et al., 2010, Mevorach et al., 2006), we performed two separate experiments, in which we stimulated the two hemispheres separately. In Experiment 1 anodal, cathodal and sham tDCS were applied to the left PPC. In Experiment 2, anodal, cathodal, and sham tDCS were applied to the right PPC.

Our prediction is that tDCS over the left and/or right PPC will modulate any polarity correspondence-driven SRC effect, because of its influence on salience processing. Specifically, we hypothesized that, if both SNARC and MARC effects were driven by the polarity correspondence principle, tDCS should modulate both effects. In contrast, if the stimulation will have differential influences on these two SRC effects, the generality of the polarity correspondence account will be undermined, in accordance with the results of Leth-Steensen and Citta (2016).