A Monitoring System for Measuring the Cognitive Cycle via a Continuous Reaction Time Task

Abstract

The cognitive cycle has been studied via cognitive architectures and by analyzing cognitive experiments. An emerging theoretical approach suggests that several automatic cognitive processes retrieve information, making it available to an internal agent, which in turn decides which information to access. Derived from this view, four phases of the cognitive cycle can be formulated and reproduced within a cognitive monitoring system. This exploratory work presents a new theory, Attention as Internal Action, and proposes a hypothesis about the relationship between an iteration of the cognitive cycle and a conscious motor action. The design of a continuous reaction time task is presented as a tool for quick cognitive evaluation. Via continuously provided user responses, the computational system behind the task adapts triggering stimuli based on the suggested hypothesis. Its software implementation was employed to assess whether a previously conducted simulation of the cognitive cycle’s time range aligned with empirical data. A control group was assigned to perform a separate simple reaction time task in a sequence of five days. The analysis showed that the experimental cognitive monitoring system produced results more closely aligned with the established understanding of the timing of the cognitive cycle than the control task did.

1. Introduction

Explaining human cognitive processing by viewing it as occurring in a cycle has been adopted by different theories [1,2,3,4]. Some of them provide understandings that is rather directed to perception by defining the cognitive schemata of the world [3]. Others explain the action knowledge that one applies by specifying schemata that generate actions [2]. On the other hand, learning types are classified in the context of the cognitive cycle, which in this case is investigated in short terms and supported with studies from neuroscience and experimental psychology [1]. These approaches are now linked to one another in order to achieve a combinative comprehension of conscious processes, schemata, and internal decision making. A combination of such understandings is presented by the theoretical approach of Attention as Action and its framework for cognitive architecture design [5,6]. It has been shown to be applicable in explaining cognitive experiences such as the crisis phenomenon [7], which occurs during the short term of the cognitive cycle [1].

The timing of the cognitive cycle has been demonstrated to be closely related to the results observed in reaction time tasks [1]. However, the LIDA Reaction Time agent designed to simulate the cognitive cycle is reported not to fully correspond to actual human subjects [1]. The main goal of this exploratory work is to investigate a new methodology for evaluating the timing of the human cognitive cycle through a continuous reaction time task, designed based on the theory Attention as Internal Action. The latter presents concepts that are systematically defined within a framework for explaining cognitive cycle iterations chaining. A theoretical principle is formulated, proposing that discrete conscious experiences depend on two factors: a complex cognitive computational system and an internal agent. Building on this principle, the Discrete Motor Execution Hypothesis is defined, offering a new perspective for conceptualizing and measuring the cognitive cycle.

The idea emerged to integrate a task design into a cognitive monitoring system, which measures reaction times across several continuous trials by adapting the target stimuli based on the motor hypothesis. To develop this idea, a set of rules is required to determine when subjects perceive a change in the environment, how and when they respond, and how their response is linked to their next perception. Corresponding to such a system of rules, a computational model can be designed that can be implemented in the software of the cognitive monitoring system, thus enabling the measurement of the duration of the human cognitive cycle.

Due to the fact that the mechanics defined by the new motor hypothesis are integrated in the continuous reaction time task, two hypothetical claims can be investigated, which can be expressed in the following way. The duration of a cognitive cycle iteration ends with the following:

• The first recorded action response signal;
• The full execution of a brief motor plan.

Based on the two claims, a mouse click response is defined in one of two ways: (1) as the act of pressing a mouse button or (2) as a brief motor plan involving both pressing and releasing the mouse button. If the cognitive monitoring system is shown to produce results closely matching the timing observed in previously conducted simulations of the cognitive cycle [1], it could provide support for the validity of the Discrete Motor Execution Hypothesis. In this way, a new bridge will be established between cognitive science and computer science. New directions for studying the cognitive cycle will be set that expand new investigations of other cognitive phenomena like conflict processing and metacognitive experiences [5,7]. These scientific efforts require a rigorous theoretical framework describing the sequential nature of the cognitive cycle.

A cognitive monitoring system was designed to integrate the continuous reaction time task in a measurement session that included an intermediate activity. The design was implemented as a software feature in an online game platform to conduct an experiment. A simple reaction time task was selected as a control task, and its results were compared with those produced by the system. Data collection took place twice: once with the control task and, separately, once with the experimental task, each over a five-day period. In total, fourteen subjects (N = 14) participated in the experiment, five of whom took part in both groups.

The content of this article initially presents established knowledge from cognitive science related to the cognitive cycle and such that is proposed by the theory of Attention as Internal Action. The computational model is also presented in the Theoretical Views Section. Next, in the Methods section, the cognitive monitoring system design is presented as well as the way in which the continuous reaction time task is integrated into it. The data analysis methods and the experimental setup are also described there. After that, the results from the experiment are presented. Lastly, in the Discussion Section, the produced data is further analyzed, and an additional hypothesis is derived and conceptually represented using the knowledge representation framework.

2. Theoretical Views

The theory behind the LIDA cognitive architecture states that consciousness is computational [8]. In contrast to this statement, a new proposal is derived from the understandings provided by the theoretical approach of Attention as Action about the cognitive cycle, learning, and automatic computational processes [5,6,7]. It is suggested that there must be two cognitive factors that lead to consciousness. If a solid theoretical framework is established about how conscious experiences occur, a computational model could be designed such that it can be integrated into a system together with a reaction time task. This way, a single iteration of the cognitive cycle [1] would not just be simulated but measurable in humans. Furthermore, if an experiment with the system is conducted and the durations align with the established understanding of the timing of the cognitive cycle [1], this would provide some evidence that the Attention as Action approach is applicable for explaining the cognitive phenomena in question.

This section first presents the theory that was formulated based on the theoretical approach of Attention as Action by explaining the semiotic technique applied in the cognitive architecture design. Secondly, it explains the hypothesis regarding the performance of motor actions within the framework of the theory by supporting it with understandings provided from established models like the Supervisory Attentional System [9], LIDA [10] and GIMA [5]. Lastly, it presents the techniques for computing the cognitive cycle via the continuous reaction time task and the computational model, which are both designed based on the formulated theory.

2.1. Attention as Internal Action

Attention as Action is still an emerging theoretical approach, but several applications now exist that demonstrate how it can be used to explain various cognitive phenomena [5,6,7]. By generally overviewing the three referenced studies, a theory can be formulated that integrates the commonly proposed concepts based on the theoretical approach. Before proceeding with the formulation of the concepts, the fundamental idea has to be explained first.

The concept of conscious experience is associated with the cognitive cycle via the Global Workspace Theory [11]. A more complete explanation of it can be provided by research on the General Theory of Behavior [12] and its application for viewing a conscious experience as a mental imagery experience [13]. Furthermore, Marks’ Action Cycle Theory [2], which is derived from his General Theory of Behavior, provides a classification and hypothetical relations between types of mental imagery experiences. The theoretical approach of Attention as Action has been applied for designing cognitive architectures of them by viewing experience types only as conscious experiences [5,6,7]. This implies that the information integrated in mental images is initially unconsciously generated by the brain and later provided for conscious access to personality. Formulating such a view would allow generally explaining two cognitive phenomena—the unconscious processing (computation) of information and conscious accessing of this information that leads to a brief conscious experience.

2.1.1. Two-Cause Internal Conjunction

The Two-cause Internal Conjunction is a principle explaining how a conscious experience occurs. In resonance with the computational theory of mind, the principle states that the brain is an incredibly complex computational system, but in order for consciousness to occur, a second factor is required. This second element operates in tight relation and in short durations of time with the computational system, which results in the occurrences of discrete conscious experiences. The complex computational system automatically processes sensory information from the environment and gathers internal information via memory processes. This corresponds to the two phases of the cognitive cycle as explained by Baars, Franklin, and others [1,14,15], namely perception and understanding, which are both unconscious and automatic. Next in the iteration of the cognitive cycle is the action selection phase, which is conscious [1]. The Two-cause Internal Conjunction principle corresponds to it; however, a key difference is that, at the start of the conscious stage, the second element begins the internal decision making phase, during which it must target a ready process in order to access information.

In terms of the theoretical approach of Attention as Action, based on which the TIC principle is formulated, the second factor is viewed as an internal agent. This is because, in correspondence to the action selection phase [1], the second element has to choose from several information products that have been produced by processes of the automatic computational system. This corresponds to the internal decision making phase, in which the internal agent targets an automatic unconscious process [7]. Making this internal decision results in the occurrence of the TIC phenomenon—a conscious experience (Figure 1).

The internal agent is now formulated as the second factor of consciousness. In the context of the targeted theoretical approach, when the automatic unconscious process accesses the provided information, the internal agent is considered to be performing an internal action [5,6,7]. The latter is coined as an act of an observation of information gathered from memory, in which the internal agent is consciously experiencing mental or perceptual imagery [2,7]. Moreover, a single internal attentional experience [5] (p. 7), as part of a continuous stream of consciousness, is also explained as a brief act of conscious learning [6]. This is following the Conscious Learning Hypothesis from the Global Workspace Theory [10] and the interaction between conscious experiences and working memory [11].

Support for this understanding can be provided by Andrade and her claim that “research should examine how working memory relates to attention and consciousness” [16]. Relations were discovered [6] that associate learning types presented in LIDA [10] and mental imagery types defined in the Action Cycle Theory [2]. An intertwining of a defined mental imagery experience with a specific type of learning is regarded as a conscious internal attentional experience [6,7]. The latter, as experienced by a personality, is now perceived as a TIC phenomenon.

2.1.2. Internal Actions and Automatic Unconscious Processes

In order to link the TIC phenomenon to the theoretical approach of Attention as Action, the concepts of internal action and of automatic unconscious processes have to be defined. They have been used for explaining crisis phenomena based on neuroscientific underpinnings on conflict processing [7], designing theoretical cognitive monitoring techniques [6], and solidifying an understanding about metacognitive experiences in cognitive architectures [5]. This shows that the concepts are implementable in a variety of comprehensions about cognitive phenomena explained in terms of events and states.

An automatic unconscious process (AUP) is complex computational brain activity that occurs without the internal agent’s awareness with the aim of retrieving or generating relevant information. After the information is gathered, the AUP emits an internal event that signifies to the internal agent that information is available (see Figure 2). In Phase 2 of the iteration of the cognitive cycle, several AUPs start to gather or process relevant information, and several of them can also emit internal events (Figure 2).

An important note is that a single iteration of the hypothetical cognitive cycle is between 260 and 390 milliseconds [1]. It starts with the perception of sensory information and continues with generating an understanding of this perception. This happens for very short durations, which explains why at the beginning of a change in the environment, the human subject cannot take immediate action. More than 100 milliseconds are required (the perception phase) in order for AUPs to gather information.

A guiding concept that serves as an underpinning for explaining how sensory information enters the computational cognitive system is the stream of incoming sensory information (SISI) [7,17]. As presented in Figure 1 and Figure 2, this cognitive element is responsible for capturing changes that occur in the continuous stream of sensory input, which is organized in a sensory information entity signified by the emission of a sensory event. The SISI is scientifically supported with the framework of Atkinson and Shiffrin, which is still influential and applicable for the formulation of new cognitive models [18]. The SISI is an abstract idea that integrates the understanding of sensory memory and how it provides event information for further processing. It is yet to be specified whether the SISI should be viewed as part of the complex computational cognitive system or as a separate (third) element. Also, several supporting cognitive architectures can be used to show the demand for the concept of the SISI. The LIDA model explains the occurrence of continuous internal and external sensory stimuli that enter the sensory memory [10]. Also, the Supervisory Attentional System model explains the sensory perceptual system that processes changes in the environment [19].

A sensory event is simply described as a change in the continuous sensory signal input that is registered by the SISI. If a sensory event produced by the SISI has to be processed in the perception phase, then there must be a specially dedicated AUP that does this. In Figure 2, the hypothesized AUP_p is depicted. This AUP is recognized as the process of perceptual associative memory [1,10,15] conceptualized in cognitive architectures produced by the theoretical approach of Attention as Action [5,6]. It is understood as a process that helps the cognitive system to quickly and non-deliberatively “recognize objects, events, and entities” [10] (p. 180). This explanation resonates with the trigger database element in the Supervisory Attentional System, the triggers of which lead to the activation of schemata [20]. That is why, corresponding to the perceptual associative AUP [5,6], the trigger AUP is formulated.

With the scientific support presented so far, the SISI and the trigger AUP can serve as underpinnings to further explain contention scheduling [19] and competition for consciousness [21]. Addressed to the prefrontal cortex [19], the Supervisory Attentional System element plays the executive role of making internal decisions. This resonates with Phase 3—the internal decision making phase, as presented in Figure 1 and Figure 2. The activated schemata [20] correspond to the AUPs that signify to the internal agent that information is ready to be provisioned (Figure 2). Only two processing AUPs that process information are depicted in Figure 2, Phase 2, but there can be more. An important note is that Phase 3 begins right after the first internal event is emitted, even though some other AUP is about to emit another internal event (Figure 2). Also, even though an AUP is conducting processing, an emission of an internal event is not always considered certain. For example, in Figure 2, AUP_p conducts processing but does not emit an internal event.

In the beginning of a conscious experience, the TIC phenomenon, a targeted AUP emits an internal event. The internal attention of the internal agent may be focused on a different AUP, so the event might not be in internal awareness. However, it remains debated whether such an event produces any internal experience content, such as mental imagery [2]. When targeted, the AUP begins to provide information that was generated computationally and without conscious awareness in Phase 2. The internal agent begins to observe this information, which results in a conscious learning experience [6]. The latter corresponds to the conscious learning as explained in LIDA [10]. In the theory formulated in this work, an internal experience corresponds to both a conscious experience type [6,10] and an imagery experience [2,7] and results from the execution of an internal action (IA). All this implies that it is plausible to name the formulated theory Attention as Internal Action (AIA).

The cognitive cycle iteration ends with the completion of the IA phase (Figure 1). However, the IA phase might be interrupted by an external (sensory) event that takes the attention away from the internal experience. If a deliberate conscious process is ongoing, during which AUPs strive to provide relevant information to satisfy the internal agent’s demand for knowledge, this process might be interrupted by a new sensory event. The SISI is continuously ongoing but some sensory events contain significantly varied sensory information that captures the subject’s external attention, which leads to diverting internal attention as well. In the deliberate conscious process, information is shared between AUPs which is referred to as AUP chaining. The interrupting sensory event breaks this chain, which leads to losing internal focus on information. This phenomenon, known as a cognitive cycle reset, is illustrated using the representation framework in the following section.

The theory of AIA uses the views of Attention as Action as underpinnings by claiming that an IA

• Occurs as a result of internal decision making;
• Is a process of conscious learning [6,11,21];
• Is performed in parallel to the information provisioning of an AUP.

Additionally, AIA formulates the concept of experience of internal attention as an occurrence of

• A perceptual or mental imagery experience [2,7];
• The TIC phenomenon.

The novelty behind the theory of AIA is the proposal that a conscious internal experience is a mental or perceptual imagery experience, while the short processes of conscious learning [6,11,21] are IAs. With this view, a big variety of concepts from cognitive science can be implemented in cognitive architecture design. Guidance is that mental or perceptual imagery types [2,22,23] are addressed in the AIA cognitive monitoring system as TIC types. Additionally, IAs have to be explained as types of conscious learning [1,10] and results of internal decision making. Now these concepts need to be implemented in an explanation about how conscious motor actions are executed. In order to do that, the semiotic technique is going to be clarified first.

2.1.3. Knowledge Representation Framework

In order to apply the theory to explain events, states, and expectations, the semiotic techniques [5,6,7] of Attention as Action have to be presented. Collectively they are going to serve as a framework for designing explanatory conceptual models of internal attention, which follow the already presented directions for cognitive monitoring via the conceptualization of scenarios of attentional experiences [6,7]. The theory of AIA has its own knowledge representation framework, which follows the rules of Attention as Action and additionally constructs its own.

By being systematically and semiotically formulated, the knowledge representation techniques from the Attention as Action models can be applied in designing computational models and digital systems for cognitive monitoring. In order to achieve this, AIA must comply with the rules of the guiding IDMA [5,7]. This means that the defined IAs and AUPs must be addressed to the General Internal Model of Attention [5]. Achieving this, the TIC phenomenon would be plausibly classified in correspondence with the imagery experience types defined in different Attention as Action models [5,6,7]. However, this does not mean that AIA must be limited to these imagery experience types. By using the semiotic techniques according to their formulated rules [5,6,7], new conscious learning processes can be linked to AUPs from cognitive science, of course by providing scientific evidence.

In the context of Attention as Action, conscious experiences have been viewed as states of internal attention [5,6,7]. Therefore, the TIC phenomenon, which corresponds to an experience of internal attention, can now be used for classifying TIC types as produced by IA and AUP types. This suggests that the TIC phenomenon should also be viewed as a short state of internal attention, which means that the semiotic techniques can also be used to depict TIC types. However, the theory of AIA views the influence of a sensory event differently.

Figure 3 depicts the knowledge representation framework of the AIA theory. It can be noticed that several consecutive cognitive cycles can be based on the same sensory information (SI). This depicts AIA’s goal to represent the idea that the sensory input signals can remain unchanged, which, of course, does not stop the continuation of the cognitive cycle. It can be argued that the SISI always produces varying information for every cognitive iteration, but the differences can be considered insignificant. This allows consecutive similar sensory events, as presented in Attention as Action [6,7], to be viewed as a single SI entity.

By following the ideas for systematically conceptualizing scenarios of internal attentional processes [6,7], in terms of AIA, the experiencing of a TIC can be viewed as such that occurs in response to

• A sensory stimulus—a new SI entity;
• The TIC from the last iteration of the cognitive cycle.

Another novelty that is proposed in this framework is the semiotic technique to depict the result from the internal decision making phase in a cognitive iteration via a dashed arrow (Figure 3). This way, the framework allows the knowledge representation of concepts like the deliberate conscious process [5], as it depicts the proceeding of several conscious experiences. Together with the representation of unconscious information transitions (Figure 3), new ideas can be provided about learning conceptualized as a proceeding of automatic memory processes as AUPs. This is represented as unconscious information transitions between the AUPs that have been chosen in several consecutive internal decision making phases. When a significantly different SI is produced by the SISI, the AUP chaining is reset, as shown in the perception TIC iteration in Figure 3. Following this idea, the theory of AIA formulates that a significantly different change in the information in the SISI leads to the occurrence of a new cognitive cycle iteration.

This framework is scientifically supported by the already established idea that the perceptual associative memory (PAM) process is targeted by perceptual learning [1,10], which corresponds to the perception IA [2,5] and the sensation IA [6,7]. In AIA, the PAM process corresponds to the trigger AUP based on the Supervisory Attentional System [9]. Most importantly for the experiment in this work, AIA’s formulations allow the Object imagery experience from the Action Cycle Theory [2] to be clearly addressed to the TIC phenomenon and the perceptual learning type [10] to be defined as an IA.

2.2. Discrete Motor Execution Hypothesis

The applications of the concepts of IAs and AUPs [5,6,7] show the internal directionality of the theoretical approach. They are useful for explaining conscious internal cognitive phenomena like experiences of mental imagery [2,12,24], metacognition [5], learning [6], conflict processing [7] and others, but can they be useful to explain conscious motor actions?

The pursuit of formulating a General Internal Model of Attention (GIMA) [5] leads to the demand for investigating IAs in the interpretational layer (layer 2), but solid support is required for the concepts in layer 1 and the physical layer. That is why this work focuses on exploring how such reliable support can be established. The model of the Supervisory Attentional System (SAS) suggests that the trigger database activates schema control units, based on which the motor effector system produces an action [19]. This corresponds to the interconnection between the Schemata and the Action modules in the Action Cycle Theory [2]. The latter has been shown to have tight relations with models of Attention as Action [5,6,7], which makes the action imagery experience linked to the motor IA [5] also known as the deed IA [5,6]. In the theory of AIA, the Action imagery experience corresponds to the TIC produced by the deed IA and the sensory motor memory AUP [5,6]. Adhering to the view of discrete conscious experiences [1], the Discrete Motor Execution Hypothesis is defined. It claims that the occurrence of a TIC—corresponding to the execution of the deed IA—is parallel to the execution of a brief, voluntary motor (external) action.

This hypothesis is supported by LIDA’s sensory motor system [10], which has been shown to be applicable in simulating the cognitive cycle [1] and in explaining smooth coping [15]. In LIDA, the sensory motor memory holds motor plan templates, which, when applied in the motor plan execution module, achieve the Online Control process [10]. The latter is a reactive mechanism in which the execution module can automatically react to sensory content directly from sensory memory via LIDA’s dorsal stream [10]. Once loaded from sensory motor memory, the motor plan is applied by the motor plan execution module, which includes triggers to motor commands [10]. This reactive mechanism can be compared to the model of the SAS [19]; however, the Online Control process lacks the modulation of the contention scheduling mechanism [10].

2.2.1. The Deed Internal Action

In terms of motor actions, the theory of AIA differs from the understandings of LIDA with its concept of the deed IA. AIA strictly follows the SAS model by claiming that each motor command is a result of contention scheduling [19], in which the SAS plays the important role of exerting “top-down control by deactivating certain schemata and activating others in the service of higher-order goals” [20]. This way the neuroscientific explanations [20,25] about the SAS model are maintained in AIA.

Following this understanding, AIA states that a single execution of a motor plan is strictly part of a cognitive cycle iteration. By stepping on the neuroscientific support [20,25], AIA constructively adds to the idea of an IA by explaining it as a process of the SAS module, which exerts top–down control [20]. Therefore, contention scheduling corresponds to AUPs competing for the attention of the internal agent and top–down control to the performance of an IA. This allows the formulation of a statement claiming that the conscious execution of a motor plan is due to the performance of a particular internal action—the deed IA.

Keeping up with the Attention as Action approach, AIA explains the execution of a motor plan as a rapid sequence of a few consecutive body action signals [5,6], occurring within a single iteration of the cognitive cycle. The body action signals are phenomena that occur in the motor cortex and evoke hand muscle responses, which take about 20 milliseconds [1,26]. A body action signal is simply the signal that is evoked from the brain to produce a single body movement. In LIDA it corresponds to a motor command produced by the Motor Plan Execution module [10,15].

In order to support striving to design a GIMA [5], AIA keeps up with the deed IA of the GIMA cognitive architecture formulated in layer 1, as presented in Figure 4. The theory of AIA has to explain how the volitional and complete execution of a single motor plan is achieved in a cognitive cycle iteration. This is achieved by, in addition to the already existing concept of unconscious information transition, formulating two supplementary flows of information. In terms of the cognitive cycle, the three information flows are defined as follows:

• Unconscious information transition: A transfer of information that occurs once in a cognitive cycle iteration between two connected AUPs [5,6].
• Continuous unconscious transmission: A constant transfer of information that proceeds throughout the cognitive cycle iteration. The internal agent cannot access the information that is being transferred.
• Continuous conscious transmission: A constant transfer of information that begins at a point in time when the internal agent is conscious. The internal agent is aware of the information that is being transferred.

Figure 4. The theory of Attention as Internal Action in terms of the General Internal Model of Attention. Acronyms: SISI—stream of incoming sensory information; SI—sensory information; SMM—sensory motor memory; IA—internal action;.

Figure 4. The theory of Attention as Internal Action in terms of the General Internal Model of Attention. Acronyms: SISI—stream of incoming sensory information; SI—sensory information; SMM—sensory motor memory; IA—internal action;.

These three information flows are depicted in Figure 4, where the lower layers of the GIMA are presented. The state change links [5], also known as links of “attentional mode switch” [6], are not depicted in this conceptual model. The theory of AIA strictly follows the rules for IA state switching presented in IDMA [7] and its inheritor—the GIMA [5].

2.2.2. The Motor Plan of the Reaction Time Task

By providing a reaction time task [27], an implementation of the designed system would require the subject to react as quickly as possible to a digital stimulus. The planned design of the task for this experiment involves a red button visualized on a dark background. The stimulus is an immediate change in the button’s color from red to green, indicating to the user to click as quickly as possible. The click is expected to be of two body action signals—pressing the index finger on the left mouse button and releasing it.

Based on the theory provided so far, the motor plan of the reaction time task can be explained with more details. As it was presented, the deed IA is an operation in the conscious stage of the cognitive cycle [1]. In the phases of the cognitive cycle iteration (Figure 1 and Figure 2), the execution of the deed IA corresponds to Phase 4. This means that the TIC phenomenon—resulting from the information provisioning by the sensory–motor memory (SMM) AUP and the execution of the deed IA—represents the conscious experience of voluntary motor plan execution. This experience can now reasonably be formulated as a brief learning experience [6] and by affirming that an IA is an act of conscious learning (Section 2.1.2), the deed IA corresponds to sensorimotor learning [10].

Based on the presented understanding of the motor plan being executed in a single cognitive cycle iteration, the duration of the IA phase can be measured. In Figure 5, the simple motor plan of clicking the mouse button is presented. The period between the two signal emissions—for pressing and releasing the finger—corresponds to the IA phase of the cognitive cycle (Figure 1). In Figure 5, this time is the period between the two dots on the timeline representation of the motor plan execution. The cognitive cycle iteration ends with the completion of the IA phase. An important note is that the computations have to consider the 20 milliseconds of “hand muscle responses” [1,26] needed for the subjects to release their finger (Figure 5). The computer system detects the user input only when the button is pressed and released but cannot detect the body action signal time for each individual. It was decided to apply the duration of 20 milliseconds in the computations for all trials and subjects.

The ideas incorporated in the motor plan execution hypothesis are used in the design of the digital cognitive task and the cognitive cycle computational model. To achieve this, the drawbacks of some reaction time tasks have to be considered.

2.3. Measuring the Human Cognitive Cycle

As a model of an artificial autonomous agent, LIDA provides an understanding of the timings of the cognitive cycle phases via simulations of the performance of a reaction time task [1]. This work aims to present a software design for a digital cognitive task intended to measure the human cognitive cycle. What is more, with the concepts formulated in Section 2.2, the IA phase can be measured. In order to keep up with the understandings of the cognitive cycle formulated in the theory of AIA, the digital cognitive task has to provide trials continuously.

2.3.1. Continuous Task Monitoring

Most of the simple reaction time tasks on the Internet, like those at https://www.arealme.com/reaction-test/en/ (accessed on 10 July 2025) and https://humanbenchmark.com/tests/reactiontime (accessed on 10 July 2025), require the user to click in order to start each trial. Also, during the trial, they do not measure the wrong clicks produced when the user clicks before the target stimulus appears. Another critique is that being conducted on the browser, the user is not entirely concentrated on the reaction time task, as some adventitious stimuli might occur.

The design for this research reaction time task was required to cope with the presented drawbacks. The concept of continuous task monitoring is considered in order to describe the method by which the subject’s cognitive cycle is being evaluated. A continuous reaction time (CRT) task is an adopted test in experimental psychology that serves as a quick tool for evaluating the alertness of subjects [28]. Used as a pre-test and a post-test for a target activity, it can provide results produced in a continuous manner. If the target CRT task produces consistent results throughout the trials, then it is plausible to claim that a specific duration of the cognitive cycle is measured.

The conduction of the target CRT task can simply be described as a series of trials, in each of which a red button appears and, at some point in time, turns green. When the subject produces a click, the green button immediately disappears. At this point, a trial is considered finished, which immediately leads to the beginning of the next trial. An important factor for accurate measuring is considered to be varying stimulus onset asynchrony [27]. In the designed CRT task, this is the time between the initial visualization of the button in red and the target stimulus—the button turning green. The gap between the disappearance of the button and the appearance of it again in red is not considered of great importance. It is decided that if a wrong click is produced before the button turns green, a gap that is a little bit longer should be introduced.

Based on the reasoning provided in this section, the targeted CRT task is required to have the following features:

• The task should be performed in full screen;
• The target stimulus should occur in an immediate manner—without animations;
• The stimulus onset asynchrony should be a multiple of the duration of the cognitive cycle;
• The stimulus onset asynchrony should be different in the trials;
• A click is considered the user input of two events, mouse button press followed by mouse button release;
• Upon a mouse button press when the button is still red, the button should immediately disappear, and negative feedback should be provided;
• Upon a click, when the button is green, the button should immediately disappear, and no feedback should be provided.

A clear but yet important note is that there must not be additional time gaps between the trials. It is considered that with the disappearance of the button, the next trial immediately starts with the pause gap preceding the button’s appearance in red again.

2.3.2. Computational Model

Considering that the feature requirements formulated in the previous section are satisfied in a digital information system, a computational model can be formulated based on several variables, equations and events. The latter are divided by perceptual events and user input events. User input events are simply defined as

• Pressing the left mouse button;
• Releasing the left mouse button.

The perceptual events are

3.

The disappearance of the button;

4.

The appearance of the button in red;

5.

The color of the button changes to green;

6.

Negative feedback text saying “Bad click!” for one second;

7.

Showing results after all the trials of the CRT task are finished.

Here are described only the variables and constants related to the computational model. The ones required for applying the computational model in the cognitive monitoring system are presented in the next section, as they are rather associated with the programming implementation. The computational model is represented by the following variables:

• G: The number of cognitive cycle iterations between event (3) and event (4);
• W: The number of cognitive cycle iterations between event (4) and event (5);
• x: The time between events (5) and (2);
• y: The time between events (5) and (1);
• C: The duration of a cognitive cycle iteration;
• A: The duration of the IA phase of a cognitive cycle iteration.

The described variables are calculated each time a trial is finished—when event (2) is produced by the user. The gap value G depends on whether the user produces a wrong click, which happens in the case when event (1) is produced before event (5). If a wrong click is produced, G is set to 7. Otherwise, upon every correctly performed trial, G is set to 5.

In order to satisfy the requirement that the stimulus onset asynchrony should be varying, a random number generating function is required for computing W. The following function is applied:

w(C) = rand(10) + add(C),

(1)

where the rand function returns a random natural number between 1 and 9. The add function returns either 2, when C is less than 200, or 0, when C is bigger than or equal to 200.

The duration of a cognitive cycle iteration (CCI) in the CRT task is the period between the two body action signals of pressing and releasing the finger (see Figure 5). This means that the travel time of the signal from the motor cortex to the hand muscle has to be considered. This forms the following equation:

C = x − m,

(2)

where m is the duration of the evocation of the hand muscle response (see Section 2.2.1). Sequential C values (CCI times) are produced in a single session with the CRT task. The CCI time includes a period, the IA phase, which is the time between the occurrence of events (1) and (2). By considering m as constant in the context of the computational model, the IA phase is calculated as follows:

A = x − y

(3)

This simple computational model depends on other variables that will be produced by the target cognitive monitoring system. These variables have to link the computations with the actual time that passes between the different events.

A question that this model answers is whether the durations of sequential CCIs are equal or close to each other. If so, then it would be considered that the implementation of this theory finds a way to measure a specific value of the CCI. What is more, a state variable of the cognitive cycle can be defined as a measurement characteristic based on the consistency of the CCIs.

2.3.3. Cognitive Cycle States

A measurement session with the implementation of the designed CRT task backed up by the computational model can provide either consistent or varying CCI results. This way, by defining a mathematical method for consistency identification, different states of the cognitive cycle can be formulated. Such states are derived from a single session with the CRT task. The main purpose of this is to be implemented as a pre- and post-test measurement tool that is used in searching for influences of an activity (the intermediate activity) in many subjects. Additionally, an important research direction is to examine the influence of subjects’ emotional state on the consistency of the results.

This work and its experiment do not focus on the evaluation of subjects’ emotional state but provides a directions for how it can be examined and how it can influence the cognitive cycle state. An emotional state is formulated as a totality that has three characteristics: stress, alertness, and emotional valence. That is why a preliminary analysis was designed that included three questionnaires:

• The Perceived Stress Scale (PSS);
• Self-Assessment Manikins (SAM);
• The Karolinska Sleepiness Scale (KSS).

The PSS involves “asking the subjects to reflect on their perceptions over general time frames (e.g., the last month)” [29]. This suggests that an experiment with the CRT task design could easily be analyzed if it is conducted in a period of a month. This way the PSS questions can be answered by every subject only at the beginning of the experiment. A suggested approach to analyze the relation between consistent CCI values and a subject’s emotional state involves applying the KSS and SAM instruments before every measurement session with the CRT task. This way, relations can be sought between consistent CCI values and the subject’s emotional state.

If a categorical variable is formulated that defines levels of consistency in CCI values, new questions can be investigated in relation to the current emotional state of the subject. Three hypotheses were formed, stating that subjects would have a lower level of consistency in CCI values when

• H1: They have a higher level of stress;
• H2: They have a higher level of arousal;
• H3: They have a higher level of sleepiness.

Testing these hypotheses would establish a solid understanding of the emotional state influence on the timing of the cognitive cycle. If solid conclusions would be drawn, emotional state could be evaluated with the CRT task to a degree of certainty without the use of measuring instruments.

3. Methods

This section begins by introducing the software implementation of the cognitive monitoring system, which included both the CRT task and the intermediate task. Next, it presents the control task and the applied methods for data analysis. Lastly, it describes the experimental setup and information related to participants.

3.1. Cognitive Monitoring System

The presented computational model and the CRT task design are recognized as a cognitive monitoring system (CMS) [6,30], which provides the novel feature of assessing the durations of a subject’s CCI. The idea was to develop a CMS that users could easily access to independently perform the CRT task with minimal effort. Before accessing the CMS, users are required to authenticate via a user management system. This enables individual results to be stored online throughout the experimental period without placing additional demands on the participants. Furthermore, the potential of using a CMS for cognitive cycle monitoring outside laboratory settings is explored. In the context of participants’ busy daily lives, such a CMS can be perceived as a quick and accessible self-monitoring tool. The design also aims to motivate participants to perform at their best without inducing stress or discouragement.

Due to the presented requirements, it was decided that the CMS would be implemented as a submodule in an already existing platform. The latter was chosen to be the online multiplayer video game Hram Light (version 4.1.18) [31], which was identified as a serious game that aims to achieve cognitive enhancement via training sessions with rational agents and humans [31,32,33]. It was assumed that the game-based format would provide subjects with intrinsic motivation to participate in the experiment. Additionally, the use of a client–server architecture enabled built-in user management features, which were utilized to ensure proper data storage and maintenance.

The host game system was developed in C++ using a proprietary software framework built upon the open-source Simple and Fast Media Library (SFML): https://www.sfml-dev.org/ (accessed on 10 July 2025). Participants were provided with a link to download a launcher that managed the installation and updates of the game client. The launcher automatically checked for the latest version and, if necessary, downloaded and installed it. All participants used the same version of the game during the experiment.

The software framework facilitated the development of the CMS as it provided established network and graphics creation utilities. Internet data transmission operations were easily developed to automatically store the results on the game server. Graphical content like the button, text feedback, the CRT task logic, and the handling of the user input events were also effortlessly settled based on the design. During the experiment, for every measurement session, each subject was able to authenticate and reliably access the cognitive monitoring feature.

Figure 6 presents a conceptual model of the CMS, integrated as a feature within the targeted game client. The client is developed using the state pattern, which is why the CMS is represented through a sequence of software states. Some states display only interface elements such as buttons and input fields (e.g., the authentication state), while others, such as the CRT task state, present dynamic content. The measurement session begins with the instructions state, leading the user to complete the CRT task. Upon completion, results are displayed, and a button initiates the intermediate activity—the fast targets task—which lasts approximately 10 s and provides immediate performance feedback. In the fifth repetition of this task, the system presents a single button that triggers transition 6, redirecting the user to perform the CRT task again (post-test). The session concludes with transition 7, marking the end of the measurement.

It is estimated that a full measurement session—including a pre-test CRT task, the fast targets activity, and a post-test CRT task—takes approximately three minutes. This short duration is considered advantageous, as it reduces the likelihood of participant fatigue or annoyance. The CRT task is of primary importance in the session, as the values it yields are central to this study’s goals.

3.1.1. The Experimental Task

A basal and important part of a computer game application is its game loop and the way it processes data in each software iteration to produce a single frame [31]. The latter can be viewed as a state of the game world or the visualized content. Different computer devices vary in performance, which means that the processing time of an identically programmed game loop iteration can differ depending on the device on which it is run. That is why software developers implement the so-called delta time variable, which stores the time that has passed since the last update of the variable in the game loop logic. Usually, this variable is processed at the beginning of each iteration; thus, put in simple terms, this is the duration for which the last game loop iteration was processed by the computer. The occurrence of an important perceptual event, like changing the color of the button to green (see Section 2.3.2), cannot happen if the next frame is not processed and rendered on the monitor.

Based on the delta time variable, a monitoring variable can be maintained in order to calculate the time passed since the last occurrence of a perceptual event. The latter is of great importance, as the theory of AIA considers the occurrence of a new, significantly different change in the continuous sensory information input to be a cognitive cycle resetting event, like the production of the SI₂ entity in the example in Figure 3. This way, the following simple flow of processing is formed for the monitoring variable (M) applied in each game loop iteration:

• Determine the elapsed time since the last arrival in this step (delta time);
• Increase the value of M with this time;
• Set M to zero if a perceptual event occurs.

In terms of the CRT task, this logic assumes that in the gap between the disappearance of the clicked button and its appearance again in red, the subject’s CCI remains with the same duration. That is why it is important that the screen in this gap remains dark without producing any perceptual stimuli. The time for which the screen remains dark is simply G multiplied by the current C. After this duration, the red button’s appearance theoretically leads to a new CCI and to a new press–release when the button turns green.

Figure 7 generally presents the implementation of the cognitive cycle monitoring based on the monitoring variable M and the delta time denoted by dt. Simply put, M measures the elapsed time from the start of the CCI. This start is dictated by the resetting (interrupting) sensory event (from Section 2.1.2). The duration of the evocation of the hand muscle response (m) is set to 20 milliseconds (see Section 2.2.2). The delta time value is approximately 16.67 milliseconds, corresponding to the game’s target frame rate of 60 frames per second. The travel time of the electrical signal resulting from the mouse button press and release is not taken into account, as it is deemed negligible. An important variable that is stochastically generated by a function based on Equation (1) (Section 2.3.2) is the number of waiting cycles while the button remains red (W). For the different CCIs, the values range from 1 to 11 and are stored on the server for each trial.

The CMS was developed to store data for each trial, including the stimulus onset time when the mouse button is pressed (y), the stimulus onset time when it is released (x), the computed duration of the cognitive cycle (C), and the count of the waiting cycles for each trial (W). The IA duration (A) can easily be calculated based on the x and y values of the trial (see Equation (3)). The analysis of the results will be addressed to the time range reported in [1], but it has to be noted that Madl, Baars, and Franklin state that the timing of the consciousness broadcast is based on “the lower limits of the times” [1] established by the underpinning experiments.

3.1.2. Intermediate Activity

A requirement was set to develop a simple intermediate activity for participants to perform after the pre-test round of the CRT task. Although not of primary importance, this activity allowed for the observation of potential differences between pre-test and post-test results. The task involved rapidly clicking on green buttons that appeared randomly in various positions on a dark screen. Each button disappeared upon being clicked and immediately reappeared elsewhere. Participants were required to complete this task five times, and the CMS feature was designed to guide users through this process.

It was decided that participants would wait five seconds at the beginning and between each round of the fast targets task (Figure 8). This task was implemented in the CMS to include 10 button appearances per round, requiring 10 correct clicks. Incorrect clicks—such as those made outside the button area—were also registered, processed, and sent to the server.

3.1.3. System Parameters

Applying the computational model, the CMS was developed to be configurable through a set of parameters. They were loaded from the server when the CMS instructions state (Figure 6) was initialized. The following parameters were implemented, by which the measurement session was determined:

• Reaction count: The number of successful trials which a user performed per single round in the CRT task;
• Muscle response time: The time required for “evoking hand muscle responses” [1];
• Iterations during pause: The count of CCIs during the time in which the user did not see the button;
• Iterations during pause after a bad click: The same as the above, but when a user clicked the button before it changed its color to green.

All of the subjects conducted the online measurement session with the same CMS parameters. Correspondingly to the order presented above, the parameters were set as follows: 10 reactions, 20 milliseconds (hand muscle response time [1,26]), 5 pause iterations after a successful trial, and 7 pause iterations following an incorrect click.

3.2. The Control Task

The experimental setup is based on pre- and post-testing with two reaction time tasks—the designed CRT task and a control one. The former was decided to be the task provided by the “A Real Me” website: https://www.arealme.com/reaction-test/en/ (accessed on 10 July 2025). The main difference in the control task compared to the CRT task lies in the technical method of recording reaction time. The control task records only the time elapsed until the subject presses the mouse button, while the CRT task also records the time of release (Figure 7). In a control task trial, the trial ends immediately after the button is pressed and the result is depicted to the user. In contrast, during a CRT task round, releasing the button is followed by a blank pause (i.e., the button disappears), after which it reappears in red. The control task has a waiting time after producing a reaction click, but it is constant (five seconds). The CRT task applies 5 or 7 CCIs for pausing but they are based on the lastly measured cognitive cycle time (Section 3.1.1), while the CRT task applies 5 or 7 CCIs for pausing, based on the most recently measured cognitive cycle time (Section 3.1.1).

An important difference is that, after pressing the button, the control task prepares the subject for the next trial by displaying a seconds timer. According to the theory of AIA, the ticking of the timer represents a significant change in the flow of sensory information—termed a SISI (Sudden and Intense Sensory Input). This suggests that an AUP reset may occur (from Section 2.1), potentially altering the eventual consistent duration of the cognitive cycle. On the other hand, the CRT task runs in full-screen mode, with no distracting elements present between trials.

3.3. Data Analysis Methods

This section describes the techniques used to organize and present the experimental data. We employed analytical techniques to generate tables and charts that facilitate meaningful observations. Further analysis is presented in the Discussion Section, where the organized results are examined using linear regression models.

3.3.1. Comparing Reaction Times

A comparative overview was conducted on the average reaction times in the control and experimental tasks. For the experimental task, it was important to consider the variable y—the measured time when the mouse button was pressed—to ensure an accurate comparison, given the technical differences in how the control task measured reaction time. A task improvement (learning) effect within a measurement session could be observed by calculating the difference between pre-test and post-test average times for each participant across the days. If the experimental group showed a smaller decrease in post-test values, this would indicate that the target stimulus of the CRT task was less predictable, and therefore the task could serve as an accurate measurement tool.

3.3.2. Hypothesis Validation

Using the reaction times from the control task, the reaction times from the CRT task, and the computed CCI durations (also from the CRT task), three datasets were obtained that could be addressed to the simulated timing of the cognitive cycle [1]. For each participant, the mean values of all pre-test rounds and all post-test rounds could be extracted and compared to the defined timing range of 260–390 milliseconds [1]. Each value would either show a negative or positive difference or fall within the defined range. In this way, an analysis could be conducted to determine which measurement most closely aligned with the defined timing.

It was expected that, among the three datasets, the reaction time results produced by the control task would most closely align with the defined timing of the cognitive cycle. This expectation stemmed from the fact that the cognitive cycle simulation was based on a simple reaction time task [1]. The Discrete Motor Execution Hypothesis asserts that the execution of a volitional motor plan is a continuous conscious process, whereas LIDA-based research does not explicitly make this claim. Consequently, the timing of the cognitive cycle—as conceptualized within the framework of the AIA theory—is expected to be slightly longer than that proposed via LIDA as the motor plan execution is after the action selection phase and is considered outside of conscious awareness [10]. The computations based on LIDA time the cognitive cycle with the end of the conscious phase (the action selection phase) [1], whereas the computational model based on the Discrete Motor Execution Hypothesis includes volitional motor executions as occurring in parallel with the IA phase.

3.3.3. Internal Action Durations

A more specific observation is that of IA phase durations. An eventual regularly occurring mode of the results was planned to be investigated. The results were planned to be manually scanned in order to find regularly occurring modes for each CRT round. For the IA phase durations where a repetitive mode was found, a table would be made where the rows were the CRT rounds and the columns corresponded to each of the ten consecutive correctly performed trials. This way, an eventual repetitive value could be observed in terms of the sequence number of each of the trials.

3.3.4. Cognitive Cycle Alteration

An expectation was that there would be some amount of sequential CCIs that would have the same values. It seems natural that at some point the subject would acquire alterations in the CCIs, but an interesting observation would be to observe when this would happen. This could be achieved by analyzing the number of waiting cycles (W) and the change occurrences in the iteration durations. An expectation was that a big change in W, e.g., from 2 to 11, would lead to a change in C.

A method for searching for variable interdependence was designed for the experimental group results. It involved examining the differences between consecutively measured CCI times and the number of CCIs applied for waiting for the target stimulus to appear (W). This method depended on the results of the variables of the CMS, the computation of which is depicted in Figure 7, Section 3.1.1. The following variables were planned to be compared:

• d_W: The count difference in W between two consecutive CRT trials;
• d_C: The difference between two consecutive CCI durations;
• u: The count of wrong clicks produced in a round with the CRT task (pre-test or post-test).

The u variable signified the degree of the user attempting to predict the target stimulus. The wrong click count was also stored and analyzed for each round with the fast targets intermediate task.

3.3.5. Cognitive Cycle Consistency

A goal was set to investigate the consistency of CCI results. An idea emerged to achieve a graphical observation by generating line charts for the CRT rounds of the reaction time results. Then, it could be observed whether the line was straight or not and how this changed over the days. Additionally, a simple technique that analyzes CCI round data was designed. The coefficient of variation was decided to be used as a signifier of consistency that could be analyzed by value ranges. This idea led to formulating the following levels of consistency based on the coefficient of variation value from each round:

• None: above 15 percent;
• Low: between 10 and 15 percent;
• Medium: between 5 and 10 percent;
• High: below 5 percent.

Calculating the coefficient of variation was not implemented in the CMS but was analyzed in the results spreadsheet. An important note is that the results produced by this technique evaluate consistency based on a one-minute conduction of the CRT task that provided ten consecutive CCI time values.

3.4. Experimental Setup

It was decided that an experimental session—whether with the control task or the experimental CRT task—would be conducted over five consecutive days. Each daily measurement session included a pre-test, an intermediate activity, and a post-test. Additionally, a rule was implemented requiring that subjects who participated in both the control and experimental conditions conducted the two sessions at least one month apart. This was intended to minimize potential learning effects from the prior experiment.

The intermediate task for the control group was required to involve reading and making simple clicks. It was decided that participants would complete basic online psychological tests to serve as the intermediate activity. A five-day plan was prepared, incorporating different psychological tests available on the web. Each test was required to take no more than ten minutes to complete.

3.4.1. Participant Selection Criteria

Fourteen participants (N = 14) were selected for the experiment. Of these, five took part in both the control and the experimental five-day sessions, three participated only in the experimental sessions, and six participated only in the control sessions. They were denoted in the following way:

• In both groups: denoted with the letters A, B, C, D and E;
• Only in the experimental group: denoted with the numbers 6–8;
• Only in the control group: denoted with the numbers 9–14.

A requirement was established to include participants with varying levels of action video game experience. Prior research has shown that simple reaction time tasks are often used to assess cognitive enhancement effects associated with playing action video games [34] and that gamers generally exhibit faster reaction times than non-gamers [35]. For this reason, both gamers and non-gamers were included in the experiment to explore potential effects of action video game play on the cognitive cycle, as conceptualized by the theory of AIA. Three participants—C, D, and 8—were identified as regular action video game players. Neither gender nor age was a selection criterion. The participants’ ages ranged from 25 to 54 years, and the sample consisted of 5 women and 9 men.

3.4.2. Expectations for Participants

The fast targets task in the targeted measurement session took only about one minute but demanded the subjects to click rapidly. That is why it was expected that the participants in the experimental group would have a decreased reaction time in the post-test. On the other hand, the designed CRT task had randomly varying time between the red button turning green in the different trials. This would prevent users from predicting the occurrence of the sensory stimulus, and therefore their reaction times would be slower.

It was anticipated that some participants would have an intuitive strive to predict the target stimulus of the tasks. This was expected to be avoided via the random generation of W values, which was the number of CCIs that the CMS applied from the appearance of the red button to the target stimulus (defined in Section 2.3.2).

4. Results

Both groups completed the control and experimental tasks at home. For the control task, participants were instructed to take screenshots or photos of their pre-test and post-test results and send them to the research team for data processing. While participants were asked not to practice the control task during the experiment, adherence could not be guaranteed. In contrast, none of the participants engaged in additional practice with the CRT task prior to or during the experiment. This was verified through the client software, which automatically transmitted online data whenever a user initiated a session with the CMS. User identities were confirmed via the email addresses used for authentication.

The results analyzed in this section are publicly available and attached to this work. Figures highlighting key aspects of the analysis are presented to facilitate the observation of important findings. This section first presents comparison between the experimental and the control task and then shows findings related to the cognitive cycle and the experimental task.

4.1. Comparing the Reaction Times

In order to investigate the developed CRT task, the method of reaction time comparison (Section 3.3.1) was applied by processing the raw data into two tables—one for the control group and one for the experimental group. Some of the participants could not attend on some of the five days in the usual plan for the experimental CRT task. This was considered a possibility to investigate how their CRT results would differ after a pause day.

Figure 9 presents the reaction time results produced by the CRT task. The table on the left shows the mean values from the ten continuous trials in each pre-test and post-test. The table on the right presents the overall mean of all pre-test and post-test values and also shows the difference calculated by subtracting the pre-test values from the post-test values. It can be observed that the sum of the differences for all of the participants is minus 0.026 s, which means that in an average measurement session, a participant improved their CRT reaction speed by 26 milliseconds. Following the idea presented in Section 3.3.1, this time value is interpreted as a task learning effect within an average measurement session.

Figure 10 presents the reaction time results produced by the simple reaction time task. Even at first glance, it is evident that the values are significantly lower compared to those in Figure 9. Moreover, the average task learning effect is approximately five times greater than that observed in the experimental group. The mean decrease in time—0.133 s—amounts to about one-third of the highest mean value (0.384 s, participant 14) and even exceeds the smallest recorded value. This suggests that the learning effect with the control task is a lot higher than the one observed in the experimental results.

By observing the results of the subjects that participated in both groups (A, B, C, D and E), it is visible that all of them had faster reaction times with the control task (Figure 9 and Figure 10). This confirms the expectation (Section 3.4.2) that the participants would have an intuitive strive to predict the occurrence of the target stimulus. It seems that the target stimulus of the control task could easily be predicted, as the sequential trials did not produce a big difference in the stimulus onset asynchrony [27].

The results of the regular action video game players (participants C, D, and 8) support the claim that action video game players have a faster reaction speed compared to non-gamers [35]. However, this is confirmed only based on the results from the experimental group—with the CRT task. If the control task results is used for analyzing gamers compared to non-gamers in this experiment, no significant difference is observed. This suggests that the CMS is applicable in analyzing effects from playing action video games.

Another observation is that, based on this small set of results, all the non-gamers (A, B, and E) had a negative general task learning effect with the CRT task (Figure 9). The action video game players C and D not only had lower reaction times than all of the others but also presented a learning effect, albeit a small one. However, in order to compare gamers to non-gamers, it is considered that an experiment with more participants is required.

4.2. Analyzing the Discrete Motor Execution Hypothesis

Thanks to the Internet data transmission to the server, the CMS sent the data of the subjects’ reaction times. For every round with the CRT task, subjects produced ten correct reactions. Two time values recorded by the CMS were of importance for every reaction. They were the values of the variables x and y from the computational model (Section 2.3.2). Additionally, for every trial, the CMS stored the W values as well.

The content in this section aims to analyze durations of cognitive cycle iterations based on the Discrete Motor Execution Hypothesis. Calculations applied based on the techniques from Section 3.3 aim to determine how closely the values from the different datasets align with the hypothesized cognitive cycle timing [1], to estimate the duration of the IA phase, to examine how and when the CCI duration was altered, and to assess the consistency of the time values within a single measurement session taken with the CRT task.

4.2.1. General Analysis of the Hypothesis

The time range of a single CCI, as presented by Madl, Baars, and Franklin [1], is between 260 and 390 milliseconds. Following the validation technique described in Section 3.3.2, it was decided that the overall means of the pre-test and post-test results for each participant would be used. Each of the three datasets—the CCI values (1), the reaction times from the CRT task (2), and the reaction times from the control task (3)—provides the overall mean values for each participant’s pre-test and post-test results. The application of this technique can be more easily observed in the following comparison of three tables.

Figure 11 presents three tables that show results from the application of the time range difference technique that validates how consistent the dataset values are with the hypothesized cognitive cycle timing. Each value is a result of one of the two following equations.

D = v − u,
v > u,

(4)

where v is the mean value of the specific pre-test or post-test round and u is the upper limit of the hypothesized time range of the cognitive cycle. When v is less than the lower limit of the cognitive cycle range, the following equation is applied:

D = v − l,
v < l,

(5)

where l is the lower limit. In this way, it can be recognized which mean values are below the lower range and which are above in Figure 11.

It can be observed that all of the control task values are below the lower limit of the cognitive cycle time range. Only one of the eleven participants showed reaction times that were within the time range. On the other hand, the results from the experimental datasets show mean time values that are more closely aligned with the hypothesized range, despite having fewer participants than the control task. An important observation is that only the regular action video game player (D) shows values in the CCI dataset that are slightly below the lower limit of the cognitive cycle time range. Overall, the results suggest that the reaction times produced by the CRT task most closely resemble the hypothesized time range.

4.2.2. Analyzing Internal Action Durations

By following the motor plan explanation for the reaction time task (Section 2.2.2), Equation (3) (Section 2.3.2) was used to calculate the IA phase for each reaction. If two or more IA phase durations that belonged to two or more consecutive CRT trials had correlating results, then this was be considered as an indication for a consistent IA phase.

Interesting results were found by analyzing the IA phase durations produced in the pre- and post-tests. They were analyzed in milliseconds, which allowed easily observing an eventual mode. A curious finding was that the results showed repetitive IA phase durations in consecutive trials in most participants. What is more, even if the consecutive CCI times differed, the IA phase duration remained the same. A great example of this occurrence is the results of participant A (Figure 12).

In Figure 13, part of the results of participant A are presented. It is visible that the value 116 milliseconds appeared as the duration of A’s IA phase quite often. Curiosity remains as to why the value almost certainly appeared in trials 3, 4, and 5, even by analyzing several days. Participant B showed a mode of 49 and 66 milliseconds, but no specific value was found as in A’s or C’s results. C showed a regularly occurring mode of 33 milliseconds. C also managed to produce a 17-millisecond IA phase. Participant D did not show a distinctive mode, and E showed regularly occurring modes but in terms of a single round. All the produced results are available in the spreadsheets dedicated to this work.

4.2.3. In Search for Alteration Interdependence

To investigate whether the cognitive cycle times were related to the change in the waiting variable (W), the method for specific analyzation from Section 3.3 was applied. The difference variables d_C and d_W were calculated using the spreadsheets presenting each of the participants’ results. For every CRT trial except the first one, d_C was presented in milliseconds and d_W as the cognitive cycle count (see more in Section 2.3.2).

Most of the results do not show interdependence between the variables. Some substantial relationship is observed in the pre-test results of participant B. As depicted in Figure 14, significant alteration interdependence is observed in trials 5, 8 and 9. In general, the two line charts appear to have similarities. This shows that in some cases the alteration in the amount of waiting for the target stimulus (W) led to alterations in the CCI time.

4.2.4. Analyzing Cognitive Cycle Consistency

The analysis of the CCI results was accomplished with plot diagrams of consecutively measured CCI time values and with the coefficient of variation technique described in Section 3.3. The aim of the former was to find flat lines, which indicated that a subject had a consistent CCI time. A curious tendency is found by observing the line charts (Figure 15). Most of them end with a single jump in the last trial in several CRT rounds. The d_W values of the last trials were analyzed. In most of the last trials where a big d_W was present, a high CCI time value was also observed. It is still curious why the big difference in the W variable did not have the same effect in the other trials as the one in the last trial. It is important to note that participants were not informed about the number of trials in the CRT task. Also, it is hard to imagine that all of the participants would have counted the consecutive trials, especially on the first day (Figure 15). The action video game players—participants C and D—also showed the same phenomenon on some of the days.

It can be seen in Figure 15 that participants quite regularly showed consistent reaction times, judging by the straight parts of the chart lines. A general heuristic observation can be made that suggests that with the passage of the experimental days, the participants had more and more consistent cognitive cycles (Figure 15). Examples of consistent cognitive cycles are presented in Figure 16. Participant C had previously been performing reaction time tasks, and that seemed to have affected the results of the CRT task. This explains the persistent values in the first round. Even the spike in the seventh trial does not show a huge difference from the other values: 33~50 milliseconds.

To analyze the consistency of the CCI times, the coefficient of variation was calculated in the spreadsheet and expressed as a percentage (see attached data for each participant). The consistency levels of the five participants that were part of both the experimental and control groups were evaluated based on the technique described in Section 3.3.5. The results evaluated for every round are presented in Figure 17. Participant C showed the lowest coefficient of variation: 3.87% on Day 2.

5. Discussion

By observing significant differences between experimental and control task results, it can be stated that the CRT task was more challenging for some participants. This is assumed to be because of the random difference between the stimulus onset asynchrony times that the CMS applied for the consecutive trials. A question remains as to whether the continuous cognitive cycle-dependent design of the CRT task was responsible for producing the consistent result or whether this depended on participants’ emotional states.

The work presented in this article focused on describing the systematically formulated explanations of the AIA theory and their implementation in the CRT task. The experiment produced two datasets containing classical reaction time data and one dataset based on the view that a reaction response constitutes the execution of a brief motor plan. The techniques for analysis were applied and results were compared.

5.1. Further Analysis

It can be confidently stated that a round of the CRT task produced reaction time values that, in a higher percentage of cases, fell within the hypothesized time range of the cognitive cycle (Figure 11, Section 4.2.1) [1]. However, the analyzed CCI durations—derived from Equation (2) and based on the interpretation of the response as a brief motor plan—showed a weaker correspondence with the hypothesized time range in comparison with the classical CRT reaction time values. An important note is that Equation (2) is based on a new idea about how the cognitive cycle can be envisioned, and even so, its values from the left table in Figure 11 are closer to the 260–390 range [1]. This suggests that the CCI values derived from Equation (2) warrant further analysis.

5.1.1. Analyzing the Computed Durations

By analyzing the values in the general comparison table (Figure 11), 25 percent of the CCI values from the left table correspond with the hypothesized time range. In contrast, only 9 percent (one out of eleven participants) from the control group fall within this range. This suggests that the computed CCI values warrant a linear regression analysis (see Figure 18).

According to the linear regression analysis, pre-test values were expected to enter the hypothesized time range by Day 6 and post-test values by Day 7. This suggests that the proposed Discrete Motor Execution Hypothesis can be used to measure the upper limit of the cognitive cycle. Over a five-day training period, with daily measurement sessions under three minutes, a quick and efficient evaluation of participants’ cognitive condition could be achieved.

By accepting the hypothesized timing of the cognitive cycle—which suggests that the motor plan for a reaction is executed after the onset of consciousness (200–280 milliseconds post-stimulus [1])—then a question arises: how could Participant D have a reaction time of just 188 milliseconds on Day 5 (Figure 9), even when performing the more demanding CRT task? This is not to mention the control task results of 152, 166, and 128 milliseconds observed on Day 1 (Figure 10). This implies that evaluating CCIs may require expanding the currently accepted cognitive cycle time range.

5.1.2. Ideas for Future Applications

In order to investigate the intangible internal experiences of human cognition, the CMS can be applied in investigating cognitive phenomena based on specific occurrences in the produced results. From a general perspective the results described in Section 4 lead to a set of phenomena that can be defined in the following way. The subject produces CCI results in a CRT round that show the following:

• An average reaction time above the hypothesized duration of the cognitive cycle;
• Consistent IA phase duration (Section 4.2.2);
• Target stimulus waiting and CCI time interdependence (Section 4.2.3);
• CCI consistency (Section 4.2.4);
• The so-called “tenth trial jump” tendency (Figure 15).

An idea is that participants that tend to have repetitively occurring values of the IA phase duration and CCI time can be regarded as being concentrated on the task and relaxed as they execute the reaction motor plan for the same duration. This can be investigated in terms of Chiksent Mihaly’s flow state to investigate loss of self-consciousness [36].

The five result phenomena, as defined above, can serve as supplementary information alongside the evaluation of CCI durations. Notably, a single CRT measurement takes approximately one minute, while a full session—including the pre-test, intermediate task, and post-test—takes around three minutes. This indicates that the CMS can play a valuable role in assessing cognitive condition. In high-stakes contexts where professionals such as surgeons, firefighters, or aircraft pilots need to be evaluated before performing critical tasks, the CMS can function as a rapid and efficient measurement tool. Additionally, the capabilities of the CMS can be extended by implementing the emotion state evaluation techniques described in Section 2.3.3. This enhancement would enable the system to provide even more detailed insights into an expert’s current cognitive condition.

5.2. Explanation for How Subjects Predict Target Stimulus

The knowledge representation technique formulated in AIA (Section 2.1.3) can be used to provide a theoretical explanation about why some participants managed to achieve extremely low reaction times in the control task. First, the applied control task had a constant waiting time before the circle appeared in red. Secondly, it had only a slightly varying target stimulus waiting time for the different trials. This allowed participants like A, 7, and 8 to achieve extremely low average reaction times, like 125 milliseconds, through the prediction of the target stimulus. If the hypothesized minimum perception phase is 80 milliseconds [1], then in this example the subject would have had only 45 milliseconds for both the understanding and action selection (conscious) phases—phases that are hypothesized to require at least 60 milliseconds [1].

Based on these observations, predicting the target stimulus seems to be the main reason why some participants managed to produce successful results in the control task. Some, like participant C, even managed to achieve this in the experimental CRT task on Day 4, when C produced a reaction time of 50 milliseconds (see attached results). By observing the other results from the round, it appears that C managed to predict the exact moment of the target stimulus. An explanation of this phenomenon can be provided by taking inspiration from the Schemata module of the Action Cycle Theory [2].

Procedural memory is shown to be tightly related to schemata mental imagery [6,7]. The theory of AIA strictly formulates that an IA corresponds to conscious learning [21] and that a TIC phenomenon type can be expressed as a mental imagery experience (Section 2.1.2). This leads to formulating schemata mental imagery [2] as a TIC phenomenon and its corresponding procedural learning [10] as an IA. It is assumed that some participants at some point have enough experiences with the reaction task that they can intuitively predict the occurrence of the target stimulus. Because intuition is tightly related to implicit learning [37,38], it is plausible to define a procedural IA as a type of cognitive process via which the internal agent consciously generates implicit knowledge. The latter is accomplished thanks to the information provisioning of the procedural memory AUP.

The theory of AIA views a procedural IA as an act of perceiving time or generating expectations of the near future in terms of a single experience of internal attention. In the case of the reaction time task, the internal agent may perform a procedural IA to predict the target stimulus. In Figure 19, two scenarios are presented via the knowledge representation framework. The upper example shows a slower reaction because of the reset caused by the new SI entity (more in Section 2.1.3). This reset is due to losing focus on the internal information because of the new attention-occupying sensory input change. The other example shows the preparedness of the internal agent for the SI₂ event achieved because of the procedural IA. In such a case, no AUP reset is present, and it is hypothesized that the internal understanding phase of the cognitive cycle [1] and the internal decision making phase (Figure 1, Section 2) take less time, resulting in a quicker reaction from the subject.

5.3. Critiques and Challenges

The CMS is an exploratory system in its initial stages of research and development. The results produced were not so easily analyzed as they were recorded in varying measurement units (seconds, milliseconds). Mean, mode, and other simple calculations could have been performed in the CMS itself, avoiding work on analyzing the experimental data.

There was a lack of instructions provided to participants, particularly during the control task. Conducting online experiments presents inherent challenges. A recommended improvement for future studies is to implement the control task directly within the CMS itself.

The consistent CCI time values that were observed in several rounds of the CRT task demand additional research methods in order for why they occur to be explained. The influence of participants’ emotional states was not investigated and the three hypotheses described in Section 2.3.3 were not tested.

The presented experiment lacked neuropsychological measurements that can support the theory behind the CMS. A plan is designed for integrating electroencephalography (EEG) devices in the CMS measurement session. This way, relations between CCI timing measured by the CMS and activations of various frequency bands could be studied [1].

No explanation is provided for the big leap tendency (Figure 15). Other mysteries that remain are the durations of the perception and the understanding (competition for consciousness) phases (Figure 2) of the cognitive cycle.

The theory is applicable in investigating the duration of the cognitive cycle based on the deed IA and the perception IA. A theoretical explanation is provided for how the procedural IA helps the internal agent to predict stimuli. However, more hypotheses and methods for investigation are required for the procedural IA and other IAs like the ones in the metacognitive layer of GIMA [5].

6. Conclusions

This work presented the formulation of the theory of Attention as Internal Action and its applicator in representing conscious and automatic unconscious processes—the knowledge representation framework. An understanding was developed that a single motor plan execution is carried out within a single cognitive cycle iteration, leading to the formulation of the Discrete Motor Execution Hypothesis. The paper described a computational model and a continuous reaction time task, both based on the hypothesis, which were implemented in a cognitive monitoring system. The analysis showed that the implemented continuous reaction time task yielded results more closely aligned with the hypothesized time range of the cognitive cycle. Additionally, a linear regression analysis of the computed cognitive cycle iteration results suggested that the developed task can be used for the evaluation of the upper time limit of the cognitive cycle.

The envisioning of a reaction response as a single brief motor plan execution raised several questions, and a new alley of investigation was discovered. The proposed design can be adapted for implementation in various systems for cognitive evaluation, making it possible to deploy a new tool for assessing individuals’ current cognitive state. In this way, the condition of professionals can be evaluated before performing critical activities such as surgical operations, firefighting missions, piloting an aircraft, or other high-stakes tasks.