Despite the general structure of the passive frame theory (PST), which is well substantiated, the concept of encapsulation of content generation (sect. 4.1 tenet in target article) seems to be inconsistent with concepts and data from other domains in cognitive psychology. My commentary will challenge it from three directions: the capacity theory and physics, perception of illusions, and higher cognitive processes such as language and planning.
Taking up the same metaphor as Morsella et al. in the target article, consciousness may be represented as an interpreter (sect. 3.3, para. 4). In the real world, an interpreter can influence both people and the debate, albeit not voluntarily, by choosing one word instead of another or by using metalinguistic signs (i.e., tones, melodies, etc.) appropriate for the output language but different from the input one (cf. the concept of semantic incommensurability: Feyerabend Reference Feyerabend1975). From a perspective taken from physics, the interpreter introduces noise into the system as she is playing a role, as stated in Heisenberg's uncertainty principle. In cognition, every process requires and consumes cognitive resources, and a certain amount of error is associated with each transformation. Because any item in the cognitive system is susceptible of being modified by near items, why should the mechanism of consciousness not be prone to this principle? By assuming that errors increase under certain circumstances (i.e., high cognitive demands, high anxiety, low activation in drowsiness, etc.), contents may be affected (and modified) while they are in the conscious field, during and after the process of generation.
If we assume that inhibition of conscious contents is different from modification of contents, we can get a satisfactory outcome. Inhibition cannot be voluntarily applied to conscious contents (as discussed by Morsella et al.), but we cannot exclude that other contents or processes may modify the contents in the conscious field. Bottom-up influences could be excluded because the model clearly substantiates that they can generate content, but what about top-down influences?
The influence of past experience on perception (Duncker Reference Duncker1939) may be helpful. If a person looks at the picture in my Figure 1, she will perceive five segments. After turning her head 90° to the left, the two letters E and F appear. From that moment on, it is not possible to perceive the segments anymore. The organized stimulus (i.e., the best guess made by the whole system, including both perception and past knowledge) hinders the initial perception in a permanent way. Along this view, Liu et al. (Reference Liu, Tzeng, Hung, Tseng and Juan2012) have demonstrated that one's own “will” can influence the conscious perception of the direction of the spinning silhouette. Therefore, if separate items in the conscious field can interact in order to change the content, the principle of encapsulation can be applied to a very large subset of items, but not to every one. Phenomena like those observed due to the mirror neuron mechanism (Gallese et al. Reference Gallese, Fadiga, Fogassi and Rizzolatti1996) may support this assertion. If encapsulation were complete, it would be really difficult to explain the relevance of certain goal-directed actions, with respect to the same motor action without a goal. While representing a food object, it does not know whether it will be eaten or thrown away. However, our cognitive system does not represent a single item alone, if other information in the environment is tightly connected to the item itself (Iacoboni et al. Reference Iacoboni, Molnar-Szakacs, Gallese, Buccino, Mazziotta and Rizzolatti2005). If a grasping action is directed towards an apple, the system conveys the message that that action is going to be performed onto the apple, even though the apple is not present in the perceptual field but hidden behind a panel. Thus, task demands are not only activated by perceptual scenery, but also they can influence low-level processes like perception and attention.
Figure 1. Does this figure depict 5 segments or 2 letters?
Although it is widely accepted that desires and beliefs cannot modify these contents, Morsella et al. state that the same is true also for high-level processes. However, their example related to Thanksgiving dinner (sect. 5, para. 10) does not foster this assertion. What they termed as “language” is just another kind of (ready-made) action plan. The stimulus may trigger a movement that could be incompatible with drinking, but this is not language unless it involves syntax, semantics, and the generation of a planned utterance. Moreover, one person could have decided to move towards the cup of chocolate and to start talking at the same time. She stops talking only while drinking, but this is just execution of a well-known pattern of action embedded in the execution of an online speech, which needs a certain level of top-down control.
The experience of insight is perfectly in accordance with the predictions made by PST as it pops up in the conscious field without any (clear) awareness of its preparation. Many tasks, instead, call for an ordered series of actions in order to obtain the desired final state (i.e., planning). In the Travelling Salesperson problem, people need to plan the shortest path between a series of subgoals (MacGregor & Ormerod Reference MacGregor and Ormerod1996). Coherently with PST, it has been demonstrated (Basso et al. Reference Basso, Bisiacchi, Cotelli and Farinello2001; Reference Basso, Lotze, Vitale, Ferreri, Bisiacchi, Olivetti Belardinelli, Rossini and Birbaumer2006) that people are aware of their intentions to change their plan during execution (due to conflicts between the initial plan and the ongoing situation). Because the number of intermediate subgoals does not affect the initial planning time, it was argued that the time before execution is used to determine the general strategy and the first movements. Thus, the actual strategy must be continuously determined during execution, and the constant planning index (Cazzato et al. Reference Cazzato, Basso, Cutini and Bisiacchi2010) justifies this assumption. Participants report to be aware of this ongoing process and that some tasks require additional effort. According to encapsulation, these participants were not able to refine their behavior during its execution, because they were aware of process outcomes only. The action-perception cycle is not involved, because planning could be made upon representations alone. Although planning may resemble mental rotation tasks, it plays a role in the approach-avoidance behavior (i.e., by determining the best path to escape), and it must be part of the behavioral repertoire of a “creature in the cave.”
You have
Access
Despite the general structure of the passive frame theory (PST), which is well substantiated, the concept of encapsulation of content generation (sect. 4.1 tenet in target article) seems to be inconsistent with concepts and data from other domains in cognitive psychology. My commentary will challenge it from three directions: the capacity theory and physics, perception of illusions, and higher cognitive processes such as language and planning.
Taking up the same metaphor as Morsella et al. in the target article, consciousness may be represented as an interpreter (sect. 3.3, para. 4). In the real world, an interpreter can influence both people and the debate, albeit not voluntarily, by choosing one word instead of another or by using metalinguistic signs (i.e., tones, melodies, etc.) appropriate for the output language but different from the input one (cf. the concept of semantic incommensurability: Feyerabend Reference Feyerabend1975). From a perspective taken from physics, the interpreter introduces noise into the system as she is playing a role, as stated in Heisenberg's uncertainty principle. In cognition, every process requires and consumes cognitive resources, and a certain amount of error is associated with each transformation. Because any item in the cognitive system is susceptible of being modified by near items, why should the mechanism of consciousness not be prone to this principle? By assuming that errors increase under certain circumstances (i.e., high cognitive demands, high anxiety, low activation in drowsiness, etc.), contents may be affected (and modified) while they are in the conscious field, during and after the process of generation.
If we assume that inhibition of conscious contents is different from modification of contents, we can get a satisfactory outcome. Inhibition cannot be voluntarily applied to conscious contents (as discussed by Morsella et al.), but we cannot exclude that other contents or processes may modify the contents in the conscious field. Bottom-up influences could be excluded because the model clearly substantiates that they can generate content, but what about top-down influences?
The influence of past experience on perception (Duncker Reference Duncker1939) may be helpful. If a person looks at the picture in my Figure 1, she will perceive five segments. After turning her head 90° to the left, the two letters E and F appear. From that moment on, it is not possible to perceive the segments anymore. The organized stimulus (i.e., the best guess made by the whole system, including both perception and past knowledge) hinders the initial perception in a permanent way. Along this view, Liu et al. (Reference Liu, Tzeng, Hung, Tseng and Juan2012) have demonstrated that one's own “will” can influence the conscious perception of the direction of the spinning silhouette. Therefore, if separate items in the conscious field can interact in order to change the content, the principle of encapsulation can be applied to a very large subset of items, but not to every one. Phenomena like those observed due to the mirror neuron mechanism (Gallese et al. Reference Gallese, Fadiga, Fogassi and Rizzolatti1996) may support this assertion. If encapsulation were complete, it would be really difficult to explain the relevance of certain goal-directed actions, with respect to the same motor action without a goal. While representing a food object, it does not know whether it will be eaten or thrown away. However, our cognitive system does not represent a single item alone, if other information in the environment is tightly connected to the item itself (Iacoboni et al. Reference Iacoboni, Molnar-Szakacs, Gallese, Buccino, Mazziotta and Rizzolatti2005). If a grasping action is directed towards an apple, the system conveys the message that that action is going to be performed onto the apple, even though the apple is not present in the perceptual field but hidden behind a panel. Thus, task demands are not only activated by perceptual scenery, but also they can influence low-level processes like perception and attention.
Figure 1. Does this figure depict 5 segments or 2 letters?
Although it is widely accepted that desires and beliefs cannot modify these contents, Morsella et al. state that the same is true also for high-level processes. However, their example related to Thanksgiving dinner (sect. 5, para. 10) does not foster this assertion. What they termed as “language” is just another kind of (ready-made) action plan. The stimulus may trigger a movement that could be incompatible with drinking, but this is not language unless it involves syntax, semantics, and the generation of a planned utterance. Moreover, one person could have decided to move towards the cup of chocolate and to start talking at the same time. She stops talking only while drinking, but this is just execution of a well-known pattern of action embedded in the execution of an online speech, which needs a certain level of top-down control.
The experience of insight is perfectly in accordance with the predictions made by PST as it pops up in the conscious field without any (clear) awareness of its preparation. Many tasks, instead, call for an ordered series of actions in order to obtain the desired final state (i.e., planning). In the Travelling Salesperson problem, people need to plan the shortest path between a series of subgoals (MacGregor & Ormerod Reference MacGregor and Ormerod1996). Coherently with PST, it has been demonstrated (Basso et al. Reference Basso, Bisiacchi, Cotelli and Farinello2001; Reference Basso, Lotze, Vitale, Ferreri, Bisiacchi, Olivetti Belardinelli, Rossini and Birbaumer2006) that people are aware of their intentions to change their plan during execution (due to conflicts between the initial plan and the ongoing situation). Because the number of intermediate subgoals does not affect the initial planning time, it was argued that the time before execution is used to determine the general strategy and the first movements. Thus, the actual strategy must be continuously determined during execution, and the constant planning index (Cazzato et al. Reference Cazzato, Basso, Cutini and Bisiacchi2010) justifies this assumption. Participants report to be aware of this ongoing process and that some tasks require additional effort. According to encapsulation, these participants were not able to refine their behavior during its execution, because they were aware of process outcomes only. The action-perception cycle is not involved, because planning could be made upon representations alone. Although planning may resemble mental rotation tasks, it plays a role in the approach-avoidance behavior (i.e., by determining the best path to escape), and it must be part of the behavioral repertoire of a “creature in the cave.”