Long-term and regular aerobic exercise –chronic aerobic exercise– has been established as a factor leading to improvements in executive control (Colcombe and Kramer, 2003; Biddle and Asare, 2011; Bherer et al., 2013; Guiney and Machado, 2013; Cox et al., 2015). Chronic aerobic exercise has also been associated with greater prefrontal (Erickson et al., 2014) and hippocampal (Leavitt et al., 2014) volumes, structures that are involved in executive functions and memory, respectively.

Padilla et al. (2017) investigated to what extent episodic memory would also benefit from chronic aerobic exercise through the improvement of executive functions. They applied a selective attention task where participants were told to remember only those items that were selectively attended. The results showed that those participants who had practiced regular aerobic exercise applied better suppression mechanisms, leading to a lower recognition rate of non-attended items. In a previous study, Padilla et al. (2013) found that active participants were able to inhibit non-relevant motor responses faster than sedentary participants using a strategic stop signal task –a highly demanding dual task-. Finally, Padilla et al. (2014) showed that active participants had a greater working memory span than sedentary participants in a dual working memory task (Unsworth et al., 2005). Taken together, these results may indicate that chronic aerobic exercise has a specific effect on the way active participants deal with highly demanding dual tasks, which requires either suppressing non-relevant information/responses or storing information while they are performing a second task. In order to explore whether the actual effect of chronic aerobic exercise in inhibitory control is related to attentional resources, it was decided to apply the “Dual-Retrieval Practice Task” (D-RP; Román et al., 2009) under the context of the Retrieval Practice paradigm (RP, Anderson et al., 1994). The D-RP task explores the role of attentional resources on exerting suppression of interference –inhibition– in memory, since it combines a working memory task with episodic memory retrieval of competing information.

Within a larger context, inhibition control has been presented as one of the core executive functions (Miyake et al., 2000). Although defining inhibition has not been an elementary task (see Harnishfeger, 1995; Nigg, 2000), some researchers have claimed the existence of a family of different types of inhibitory processes (e.g., Harnishfeger, 1995; Hasher et al., 1999; Nigg, 2000; Friedman and Miyake, 2004). One of the features that has been argued to be crucial in distinguishing between different types of inhibitory tasks is the required level of executive control involved. According to several authors (Harnishfeger, 1995; Nigg, 2000; Friedman and Miyake, 2004; Andrés et al., 2008), inhibitory tasks can be classified into a continuum from very controlled to very automatic, and this also applies to RP. It must be taken into account that these authors (Harnishfeger, 1995; Nigg, 2000; Friedman and Miyake, 2004; Andrés et al., 2008) assume that the type of inhibition that requires a high executive control indirectly demands a high level of attentional resources. Those resources will allow non-relevant information or processes to be filtered out, keep mental representations active while mental operations are being exerted, but also to allocate attention to different tasks if necessary. Thus, controlled inhibition requires a process of collaboration along with working memory (Hasher and Zacks, 1988; Bjork, 1989; Redick et al., 2011).

Returning to the RP paradigm, it is meant to explore forgetting when interference is present. When only certain items from defined list need to be recalled, forgetting the irrelevant items can be a necessary and adaptive process. Under these circumstances, it has been shown that intentionally recalling information may lead to forgetting other information (Wimber et al., 2015). In other words, the very act of remembering information may be one of the major reasons why we forget other information.

The Ordinary Retrieval Practice task (O-RP; Anderson et al., 1994) consists of three phases. First, in the study phase, several lists of category-exemplars (FRUIT-Orange; REPTILE-Snake) are studied. Each category contains multiple exemplars. Second, in the retrieval practice phase, only 50% of the exemplars from one half of the categories must be retrieved with the aid of a cue (FRUIT-Or____). Each exemplar is retrieved a fixed number of times during this phase. Finally, in the recognition phase, all exemplars are presented one-by-one and participants must decide if the exemplar was studied or not in the study phase (see Hicks and Starns, 2004; Gómez-Ariza et al., 2005; Bajo et al., 2006; for recognition based RIF).

The exemplars that were practiced during the retrieval practice phase are considered Rp+. The exemplars that were not retrieved but belonged to the categories practiced in the retrieval phase are called Rp−. Finally, the exemplars that were not retrieved and did not belong to the categories practiced in the retrieval phase are named Nrp. The typical key finding is that the non-practiced (Rp−) items are later less recognized than the control items (Nrp). One of the most plausible explanations is that this retrieval induced forgetting (RIF) effect is the consequence of an inhibitory mechanism that functions at the retrieval phase to reduce the activation of competitive non-target items (Rp−) that interfere with the retrieval of target items (Rp+, Anderson et al., 1994). As a consequence, later in the recognition phase, Rp− items are more difficult to access than Nrp items even though both were equally non-practiced (see Murayama et al., 2014 for a review).

Further studies have explored the extent to which the RIF effect may be modulated by executive resources (Román et al., 2009; Aslan and Bäuml, 2011; Ortega et al., 2012). In this way, Román et al. (2009) designed the “Dual-Retrieval Practice Task” (D-RP) combining a working memory task (remembering five digits or updating a sequence of three numbers) with the retrieval of the previously studied exemplars. They showed that in the dual condition –digit retrieval/updating plus exemplar retrieval–, the RIF effect was absent given that accuracy and response times were similar for Rp− and Nrp items in the recognition phase. This result suggested that some attentional resources are involved in the inhibition of memories at the retrieval phase, despite the lack of intentionality. Subsequently, Ortega et al. (2012) applied the D-RP task in a sample of young and older adults to investigate the effects of aging on the RIF effect. The results showed that there was a RIF effect in the young group when the updating task consisted of three digits, but not when consisting of five digits, confirming the idea that memory inhibition is dependent on executive control. Moreover, RIF was not observed in the older adult group even for the three digits updating D-RP task, revealing that the reduction in executive resources typically observed in aging led to the disappearance of the RIF effect in the D-RP task. Finally, in line with the idea that RIF involved some executive control, Aslan and Bäuml (2011) showed a positive correlation between RIF and working memory capacity in younger adults.

Until now, several studies looking at individual differences in RIF have concentrated on populations that were supposed to show a reduced forgetting effect observed in the RP task (Amir et al., 2009; Soriano et al., 2009; Law et al., 2012; Saunders, 2012; Storm and Jobe, 2012). However, we wanted to investigate whether the RIF effect may be potentiated, especially by a factor (aerobic exercise) that has recently received increased attention in psychological research.

If the type of inhibition involved in the D-RP task can be modulated by executive resources, as has been shown in previous studies (Román et al., 2009; Ortega et al., 2012), then the D-RP task should be sensitive to physical activity, a factor that has been shown to improve executive functions. The hypothesis of the current study is that active participants will display additional executive resources compared to sedentary participants, and this will allow them to deal with the interference induced by the dual-working memory task whereas they suppress the Rp− items. In other words, we predict that physically active participants will display a RIF effect despite being in a dual-task situation, whereas sedentary participants will not.

The aim of this study was to investigate whether the inhibitory processes involved in the retrieval of competitive elements from memory might be positively modulated by chronic aerobic exercise through attentional resources. The main prediction was that physically active participants would show RIF under a high demanding executive task –D-Rp−, whereas sedentary participants would not.

The results first indicated that active and sedentary participants did not differ in factors that could have improved executive control in active participants such as bilingualism, education, crystallized and fluid intelligence or age. This equality between sedentary and active participants on non-manipulated and potentially confounding variables makes potential group differences in D-RP statistically valid.

There remains, however, some putative uncontrolled variables that might affect RIF. For example, some studies have described self-control and self-regulation differences between active and sedentary participants (Audiffren and André, 2015; Diamond and Ling, 2016, 2018) and this might mediate the observed RIF differences between groups. This would be an empirical question that could be investigated in future studies.

As usually observed in the D-RP task (Ortega et al., 2012), participants did not differ in performance during the retrieval-practice phase. However, active and sedentary groups showed different patterns of recognition. First, while active participants showed a significant RIF effect, the sedentary group did not. Active participants recognized significantly fewer Rp− items than Nrp. Also, active participants recognized fewer Rp− than sedentary participants. Since previous studies have revealed that physically active participants showed improved executive functions (Guiney and Machado, 2013; Padilla et al., 2013, 2014; Pérez et al., 2014), it is likely that active participants were able to suppress Rp− memory traces at retrieval practice despite the concurrent working memory load (updating task). The absence of differences in the digit-recall task –when performed before the dual task– and the similar levels of the Nrp recognition suggest that physical activity specifically modulates the ability to inhibit competing information when executive control is taxed by dual tasking. Therefore, in line with previous studies (Padilla et al., 2013, 2014, 2017; Pérez et al., 2014) our results indicate that physical activity selectively affects inhibitory control through attentional resources.

It is worth noting that the pattern of results revealed by sedentary participants in forgetting was as expected in young adults. Ortega et al. (2012) showed that young participants, when carrying the D-RP task with 5 digits, had an absence of RIF effect. The authors explained this finding by the fact that divided attention caused such a reduction in attentional resources that active suppression of Rp− interference during retrieval could not be undertaken. Similarly, the sedentary participants from the current study did not have enough attentional resources to suppress Rp− interference while dealing with the working memory –updating– task. It can be argued that the chronic practice of physical activity has led to an increase in attentional resources in the active participants, making it possible for them to suppress the interference from the Rp− at the retrieval phase.

Alternatively, the differences in the recognition of the Rp+ items for the sedentary and active group could result from the fact that they were presented in the final part of the recognition phase and after Rp− and Nrp items had been presented. Thus, during the first block of the recognition phase, the retrieval of the Rp− may cause inhibition of the competing Rp+, and this in turn will produce lower recognition of the Rp+ items in the recognition block. Considering sedentary participants demonstrate, in general, less inhibitory capacities, the RP+ items will not be inhibited during the first recognition phase and therefore facilitation is still present for the practiced items. This cost-benefit interplay has been previously illustrated and discussed (Anderson and Levy, 2007).

The data suggest that greater attentional resources in active participants result in a greater ability to suppress interference when working memory is loaded in the dual-RP updating task. This leads to the understanding that inhibition during retrieval is susceptible to modification by aerobic exercise through the mediation of attentional resources that facilitates divided attention between two tasks. The previous results (Padilla et al., 2013, 2014, 2017; Pérez et al., 2014) were replicated and extended by showing that a long-term routine of physical activity is associated with a improved ability to deal with a demanding dual task such as the D-RP. It would be interesting to corroborate the results with an aerobic exercise intervention, since cross-sectional studies do not allow establishing causal relationships.

In this regard, long-term cardiovascular exercise is accepted as an intervention that benefits cognition since early adulthood. Its effect on attentional resources is reflected in different cognitive functions. These findings may convince policymakers of the importance of exercise and promote the lifelong practice of cardiovascular exercise.

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher. Requests to access the datasets should be directed to Dr. Concepcion Padilla at cfp31@medschl.cam.ac.uk.

TB, PA, and CP designed the study. CP wrote the paper, analyzed the data, recruited and applied the tests to the participants. PA and TB gave advice and ideas on the analyses and the interpretation of results and helped on the writing of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.