Limits...
Neuronal Correlates of Cognitive Control during Gaming Revealed by Near-Infrared Spectroscopy.

Witte M, Ninaus M, Kober SE, Neuper C, Wood G - PLoS ONE (2015)

Bottom Line: We found an increased change of oxygenated and deoxygenated hemoglobin during LEARN covering broad areas over right frontal, central and parietal cortex.Opposed to this, hemoglobin changes were less pronounced for RANDOM and APPLY.Along with the findings that fewer objects were caught during LEARN but stimulus-response mappings were successfully identified, we attribute the higher activations to an increased cognitive load when extracting an unknown mapping.

View Article: PubMed Central - PubMed

Affiliation: Department of Psychology, University of Graz, Universitaetsplatz 2, 8010 Graz, Austria; BioTechMed Graz, Graz, Austria.

ABSTRACT
In everyday life we quickly build and maintain associations between stimuli and behavioral responses. This is governed by rules of varying complexity and past studies have identified an underlying fronto-parietal network involved in cognitive control processes. However, there is only limited knowledge about the neuronal activations during more natural settings like game playing. We thus assessed whether near-infrared spectroscopy recordings can reflect different demands on cognitive control during a simple game playing task. Sixteen healthy participants had to catch falling objects by pressing computer keys. These objects either fell randomly (RANDOM task), according to a known stimulus-response mapping applied by players (APPLY task) or according to a stimulus-response mapping that had to be learned (LEARN task). We found an increased change of oxygenated and deoxygenated hemoglobin during LEARN covering broad areas over right frontal, central and parietal cortex. Opposed to this, hemoglobin changes were less pronounced for RANDOM and APPLY. Along with the findings that fewer objects were caught during LEARN but stimulus-response mappings were successfully identified, we attribute the higher activations to an increased cognitive load when extracting an unknown mapping. This study therefore demonstrates a neuronal marker of cognitive control during gaming revealed by near-infrared spectroscopy recordings.

No MeSH data available.


Related in: MedlinePlus

Topographical patterns.(A) Grand average topographies for the time period 6 to 12 seconds. Numbers indicate channels as shown in Fig 1. (B) Mapping of the activations (numbers again represent channels) to a two-dimensional space by multidimensional scaling. We used pairwise correlation between channels as input to an iterative procedure based on the smacof algorithm (for details see Methods). Channels with a high correlation will cluster together, thus revealing the structure of the whole network (stress values for RANDOM, LEARN and APPLY: 1.3, 1.0 and 0.6). (C) Same data as in (B) mapped into a common space (stress value: 0.003). Note that the resulting configuration dimensions are without a unit and lower stress values indicate a better fit.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4526694&req=5

pone.0134816.g003: Topographical patterns.(A) Grand average topographies for the time period 6 to 12 seconds. Numbers indicate channels as shown in Fig 1. (B) Mapping of the activations (numbers again represent channels) to a two-dimensional space by multidimensional scaling. We used pairwise correlation between channels as input to an iterative procedure based on the smacof algorithm (for details see Methods). Channels with a high correlation will cluster together, thus revealing the structure of the whole network (stress values for RANDOM, LEARN and APPLY: 1.3, 1.0 and 0.6). (C) Same data as in (B) mapped into a common space (stress value: 0.003). Note that the resulting configuration dimensions are without a unit and lower stress values indicate a better fit.

Mentions: Topographical plots of the average Hb-differences described statistically above corroborated an increased activation during LEARN over a broad cortical region (Fig 3A). The main focus was located over central sites extending to parietal and prefrontal regions. Similar albeit weaker changes were found for the RANDOM task, with a focus over primary sensorimotor areas. In contrast, the maximum activations during the APPLY task were mostly confined to posterior parietal and dorsolateral prefrontal areas. Because these average activation plots can only represent a first hint on the main areas involved, multidimensional scaling was used to visualize the different networks both within and across tasks according to their neuronal similarity [44]. We decided to use correlation over time as an index of functional connectivity here. Similar approaches have been applied for example to identify language networks during speech comprehension [48]. Multidimensional scaling optimized the positions of single channels in a low-dimensional space to represent the overall ‘similarity structure’ of a higher dimensional space [49]. Fig 3B displays the resulting two-dimensional coordinates that represent the similarities based on the correlation between pairs of channels. The LEARN task was associated with basically one extended cluster. In contrast, two separate clusters covering mostly pre- and post-central or fronto-parietal regions were identified for the RANDOM task. For the APPLY task, channels over primary sensorimotor areas were grouped together and slightly stood out from the remaining channels. Within this second larger cluster, a subset of fronto-parietal channels (channels 1, 3, 16, 24) could be observed. In addition to these clusters within a given task, we also pooled the correlation matrices of all three tasks as input to multidimensional scaling. Besides the identification of fine patterns, this provided a global comparison of the connectivity patterns across tasks (Fig 3C). The analysis revealed a high similarity between the LEARN and APPLY tasks, while the RANDOM task showed an overall different configuration.


Neuronal Correlates of Cognitive Control during Gaming Revealed by Near-Infrared Spectroscopy.

Witte M, Ninaus M, Kober SE, Neuper C, Wood G - PLoS ONE (2015)

Topographical patterns.(A) Grand average topographies for the time period 6 to 12 seconds. Numbers indicate channels as shown in Fig 1. (B) Mapping of the activations (numbers again represent channels) to a two-dimensional space by multidimensional scaling. We used pairwise correlation between channels as input to an iterative procedure based on the smacof algorithm (for details see Methods). Channels with a high correlation will cluster together, thus revealing the structure of the whole network (stress values for RANDOM, LEARN and APPLY: 1.3, 1.0 and 0.6). (C) Same data as in (B) mapped into a common space (stress value: 0.003). Note that the resulting configuration dimensions are without a unit and lower stress values indicate a better fit.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4526694&req=5

pone.0134816.g003: Topographical patterns.(A) Grand average topographies for the time period 6 to 12 seconds. Numbers indicate channels as shown in Fig 1. (B) Mapping of the activations (numbers again represent channels) to a two-dimensional space by multidimensional scaling. We used pairwise correlation between channels as input to an iterative procedure based on the smacof algorithm (for details see Methods). Channels with a high correlation will cluster together, thus revealing the structure of the whole network (stress values for RANDOM, LEARN and APPLY: 1.3, 1.0 and 0.6). (C) Same data as in (B) mapped into a common space (stress value: 0.003). Note that the resulting configuration dimensions are without a unit and lower stress values indicate a better fit.
Mentions: Topographical plots of the average Hb-differences described statistically above corroborated an increased activation during LEARN over a broad cortical region (Fig 3A). The main focus was located over central sites extending to parietal and prefrontal regions. Similar albeit weaker changes were found for the RANDOM task, with a focus over primary sensorimotor areas. In contrast, the maximum activations during the APPLY task were mostly confined to posterior parietal and dorsolateral prefrontal areas. Because these average activation plots can only represent a first hint on the main areas involved, multidimensional scaling was used to visualize the different networks both within and across tasks according to their neuronal similarity [44]. We decided to use correlation over time as an index of functional connectivity here. Similar approaches have been applied for example to identify language networks during speech comprehension [48]. Multidimensional scaling optimized the positions of single channels in a low-dimensional space to represent the overall ‘similarity structure’ of a higher dimensional space [49]. Fig 3B displays the resulting two-dimensional coordinates that represent the similarities based on the correlation between pairs of channels. The LEARN task was associated with basically one extended cluster. In contrast, two separate clusters covering mostly pre- and post-central or fronto-parietal regions were identified for the RANDOM task. For the APPLY task, channels over primary sensorimotor areas were grouped together and slightly stood out from the remaining channels. Within this second larger cluster, a subset of fronto-parietal channels (channels 1, 3, 16, 24) could be observed. In addition to these clusters within a given task, we also pooled the correlation matrices of all three tasks as input to multidimensional scaling. Besides the identification of fine patterns, this provided a global comparison of the connectivity patterns across tasks (Fig 3C). The analysis revealed a high similarity between the LEARN and APPLY tasks, while the RANDOM task showed an overall different configuration.

Bottom Line: We found an increased change of oxygenated and deoxygenated hemoglobin during LEARN covering broad areas over right frontal, central and parietal cortex.Opposed to this, hemoglobin changes were less pronounced for RANDOM and APPLY.Along with the findings that fewer objects were caught during LEARN but stimulus-response mappings were successfully identified, we attribute the higher activations to an increased cognitive load when extracting an unknown mapping.

View Article: PubMed Central - PubMed

Affiliation: Department of Psychology, University of Graz, Universitaetsplatz 2, 8010 Graz, Austria; BioTechMed Graz, Graz, Austria.

ABSTRACT
In everyday life we quickly build and maintain associations between stimuli and behavioral responses. This is governed by rules of varying complexity and past studies have identified an underlying fronto-parietal network involved in cognitive control processes. However, there is only limited knowledge about the neuronal activations during more natural settings like game playing. We thus assessed whether near-infrared spectroscopy recordings can reflect different demands on cognitive control during a simple game playing task. Sixteen healthy participants had to catch falling objects by pressing computer keys. These objects either fell randomly (RANDOM task), according to a known stimulus-response mapping applied by players (APPLY task) or according to a stimulus-response mapping that had to be learned (LEARN task). We found an increased change of oxygenated and deoxygenated hemoglobin during LEARN covering broad areas over right frontal, central and parietal cortex. Opposed to this, hemoglobin changes were less pronounced for RANDOM and APPLY. Along with the findings that fewer objects were caught during LEARN but stimulus-response mappings were successfully identified, we attribute the higher activations to an increased cognitive load when extracting an unknown mapping. This study therefore demonstrates a neuronal marker of cognitive control during gaming revealed by near-infrared spectroscopy recordings.

No MeSH data available.


Related in: MedlinePlus