We were surprised to read that Chaz Firestone and Brian Scholl (F&S) consider cognition and perception as purely psychological mechanisms. Like the vast majority of professional neuroscientists worldwide, we consider that cognitions and perceptions are governed by specific patterns of electrical and chemical activity in the brain, and are thus essentially physiological phenomena. We therefore consider that cognitions and perceptions are embodied. Accordingly, we propose that brain activity best explains top-down mechanisms.
First, double-task paradigms, as proposed by Schwartz et al. (Reference Schwartz, Vuilleumier, Hutton, Maravita, Dolan and Driver2005), allow testing focused attention at the centre of a screen (monitored by eye-tracking), in which attentional load at the centre while detecting various shapes and colours in a continuous stream of letters – L's and T's – can be varied simply by changing the instructions.
Simultaneously, irrelevant stimuli are presented at the periphery to activate specific areas of the ventral visual pathway (e.g., visual area 4 – V4) independently of the main attentional task.
Double-tasking enables the determination of whether certain top-down effects are related exclusively (or not) to intentional control – for example, during a directed attentional task, as F&S demonstrate. In our studies, we used divided attention in a double-task design that simultaneously varied the attentional load centrally and the perception of irrelevant coloured stimuli at the periphery (Desseilles et al. Reference Desseilles, Balteau, Sterpenich, Dang-Vu, Darsaud, Vandewalle, Albouy, Salmon, Peters, Schmidt, Schabus, Gais, Degueldre, Phillips, Luxen, Ansseau, Maquet and Schwartz2009; Reference Desseilles, Schwartz, Dang-Vu, Sterpenich, Ansseau, Maquet and Phillips2011).
Second, during our tasks we measured blood-oxygen-level dependant (BOLD) signals elicited by neuronal activity throughout the brain. Signal modulations resulting from task effects reflect changes in regional activity, and hence changes in brain perceptions: in other words, sensations.
This strategy differs considerably from the standard method of directly asking subjects about how their perceptions vary with performed tasks. Verbal self-reports involve several biases, such as social desirability bias. The result is a subjective, post hoc impression of what was meant to be an objective measure of perception. In contrast, brain imaging provides a practically real-time measure of brain perception that avoids the desirability bias as well as subjective self-reports and consequent interpretability problems, for both subjects and researchers. Accordingly, we consider perception a neuronal activity that can be measured objectively. The issue of whether the BOLD signal reflects neuronal activity alone does not affect this approach. In fact, we agree with the authors that between the input (subject stimulation) and the output (subject's self-report), if the internal representations are psychological only, it would be impossible to pinpoint the top-down influence of cognition on perception. We therefore propose that brain activity measures should be used to unravel the mechanisms of top-down dynamics (Desseilles et al. Reference Desseilles, Schwartz, Dang-Vu, Sterpenich, Ansseau, Maquet and Phillips2011). Uncontrollable and highly variable phenomena call for precise measures.
Third, brain activity modelling is performed in a series of steps. The first step is to identify brain areas that show significant activity modulation by the task (Friston et al. Reference Friston, Ashburner, Kiebel, Nichols and Penny2007). The second step is to map functional connectivity in terms of psychophysiological interactions (Gitelman et al. Reference Gitelman, Penny, Ashburner and Friston2003). Although this approach does not allow determining causality between areas, it tests interactions between psychological factors (here, attentional load) and physiological factors (here, brain activity) (Desseilles et al. Reference Desseilles, Balteau, Sterpenich, Dang-Vu, Darsaud, Vandewalle, Albouy, Salmon, Peters, Schmidt, Schabus, Gais, Degueldre, Phillips, Luxen, Ansseau, Maquet and Schwartz2009). After these exploratory steps, causal relationships between identified sets of brain regions can be determined by generative mechanistic modelling of brain responses, also called dynamic causal modelling (Friston et al. Reference Friston, Harrison and Penny2003). In this framework, concurrent hypotheses on brain functioning are expressed by different connectivity models, and the optimal model for a population can be identified from the subjects' brain activity (Desseilles et al. Reference Desseilles, Schwartz, Dang-Vu, Sterpenich, Ansseau, Maquet and Phillips2011; Penny et al. Reference Penny, Stephan, Daunizeau, Rosa, Friston, Schofield and Leff2010).
Fourth, the cognitive charge used in our tasks depended on the different populations that participated in the studies, including healthy controls and depressed patients (Desseilles et al. Reference Desseilles, Balteau, Sterpenich, Dang-Vu, Darsaud, Vandewalle, Albouy, Salmon, Peters, Schmidt, Schabus, Gais, Degueldre, Phillips, Luxen, Ansseau, Maquet and Schwartz2009; Reference Desseilles, Schwartz, Dang-Vu, Sterpenich, Ansseau, Maquet and Phillips2011). Importantly, the inclusion of pathophysiology sheds new light on top-down connectivity, and investigations of perception and cognition should consider mental illnesses that directly affect cognition. Similarly, a paradigmatic shift has taken place, from psychological studies of personality to clinical case studies, for instance, by Phineas Gage. In view of all of the above, we argue that brain activity modelling should be the method of choice for unravelling top-down dynamics.
We were surprised to read that Chaz Firestone and Brian Scholl (F&S) consider cognition and perception as purely psychological mechanisms. Like the vast majority of professional neuroscientists worldwide, we consider that cognitions and perceptions are governed by specific patterns of electrical and chemical activity in the brain, and are thus essentially physiological phenomena. We therefore consider that cognitions and perceptions are embodied. Accordingly, we propose that brain activity best explains top-down mechanisms.
First, double-task paradigms, as proposed by Schwartz et al. (Reference Schwartz, Vuilleumier, Hutton, Maravita, Dolan and Driver2005), allow testing focused attention at the centre of a screen (monitored by eye-tracking), in which attentional load at the centre while detecting various shapes and colours in a continuous stream of letters – L's and T's – can be varied simply by changing the instructions.
Simultaneously, irrelevant stimuli are presented at the periphery to activate specific areas of the ventral visual pathway (e.g., visual area 4 – V4) independently of the main attentional task.
Double-tasking enables the determination of whether certain top-down effects are related exclusively (or not) to intentional control – for example, during a directed attentional task, as F&S demonstrate. In our studies, we used divided attention in a double-task design that simultaneously varied the attentional load centrally and the perception of irrelevant coloured stimuli at the periphery (Desseilles et al. Reference Desseilles, Balteau, Sterpenich, Dang-Vu, Darsaud, Vandewalle, Albouy, Salmon, Peters, Schmidt, Schabus, Gais, Degueldre, Phillips, Luxen, Ansseau, Maquet and Schwartz2009; Reference Desseilles, Schwartz, Dang-Vu, Sterpenich, Ansseau, Maquet and Phillips2011).
Second, during our tasks we measured blood-oxygen-level dependant (BOLD) signals elicited by neuronal activity throughout the brain. Signal modulations resulting from task effects reflect changes in regional activity, and hence changes in brain perceptions: in other words, sensations.
This strategy differs considerably from the standard method of directly asking subjects about how their perceptions vary with performed tasks. Verbal self-reports involve several biases, such as social desirability bias. The result is a subjective, post hoc impression of what was meant to be an objective measure of perception. In contrast, brain imaging provides a practically real-time measure of brain perception that avoids the desirability bias as well as subjective self-reports and consequent interpretability problems, for both subjects and researchers. Accordingly, we consider perception a neuronal activity that can be measured objectively. The issue of whether the BOLD signal reflects neuronal activity alone does not affect this approach. In fact, we agree with the authors that between the input (subject stimulation) and the output (subject's self-report), if the internal representations are psychological only, it would be impossible to pinpoint the top-down influence of cognition on perception. We therefore propose that brain activity measures should be used to unravel the mechanisms of top-down dynamics (Desseilles et al. Reference Desseilles, Schwartz, Dang-Vu, Sterpenich, Ansseau, Maquet and Phillips2011). Uncontrollable and highly variable phenomena call for precise measures.
Third, brain activity modelling is performed in a series of steps. The first step is to identify brain areas that show significant activity modulation by the task (Friston et al. Reference Friston, Ashburner, Kiebel, Nichols and Penny2007). The second step is to map functional connectivity in terms of psychophysiological interactions (Gitelman et al. Reference Gitelman, Penny, Ashburner and Friston2003). Although this approach does not allow determining causality between areas, it tests interactions between psychological factors (here, attentional load) and physiological factors (here, brain activity) (Desseilles et al. Reference Desseilles, Balteau, Sterpenich, Dang-Vu, Darsaud, Vandewalle, Albouy, Salmon, Peters, Schmidt, Schabus, Gais, Degueldre, Phillips, Luxen, Ansseau, Maquet and Schwartz2009). After these exploratory steps, causal relationships between identified sets of brain regions can be determined by generative mechanistic modelling of brain responses, also called dynamic causal modelling (Friston et al. Reference Friston, Harrison and Penny2003). In this framework, concurrent hypotheses on brain functioning are expressed by different connectivity models, and the optimal model for a population can be identified from the subjects' brain activity (Desseilles et al. Reference Desseilles, Schwartz, Dang-Vu, Sterpenich, Ansseau, Maquet and Phillips2011; Penny et al. Reference Penny, Stephan, Daunizeau, Rosa, Friston, Schofield and Leff2010).
Fourth, the cognitive charge used in our tasks depended on the different populations that participated in the studies, including healthy controls and depressed patients (Desseilles et al. Reference Desseilles, Balteau, Sterpenich, Dang-Vu, Darsaud, Vandewalle, Albouy, Salmon, Peters, Schmidt, Schabus, Gais, Degueldre, Phillips, Luxen, Ansseau, Maquet and Schwartz2009; Reference Desseilles, Schwartz, Dang-Vu, Sterpenich, Ansseau, Maquet and Phillips2011). Importantly, the inclusion of pathophysiology sheds new light on top-down connectivity, and investigations of perception and cognition should consider mental illnesses that directly affect cognition. Similarly, a paradigmatic shift has taken place, from psychological studies of personality to clinical case studies, for instance, by Phineas Gage. In view of all of the above, we argue that brain activity modelling should be the method of choice for unravelling top-down dynamics.