Introduction

Approximately 20% of adults identify themselves as chronic procrastinators, with as many as 85-90% of students involved in academic procrastination (Procrastination Statistics). Procrastination results in increased stress, guilt, and a sense of decreased achievement, which in turn can contribute to mental health problems such as anxiety. Furthermore, procrastination often leads to a lower quality of work and missed deadlines, which negatively affects career advancement and personal goals, further damaging self-esteem. There are various causes of procrastination, namely fear of failure, perfectionism, and a lack of motivation—procrastination fundamentally stems from an inability to manage negative emotions. Management techniques like creating a schedule or study plan may even exacerbate the problem if the person has no discipline to commit in the first place.

As such, addressing procrastination requires a deep understanding of cognitive-behavioral techniques to alter underlying thought patterns that contribute to delaying tasks and create a mental habit of discipline. However, existing scheduling methods often struggle to adapt to changing conditions, and existing study apps often lack the personalization and adaptability needed to truly optimize productivity and accommodate diverse learning styles and preferences. Furthermore, most focus on building extrinsic motivation, such as using rewards or punishments to encourage task completion. While these methods can be effective in the short term, they do not foster intrinsic motivation, which is crucial for long-term engagement and sustainable productivity. In fact, relying solely on extrinsic motivators diminishes the inherent satisfaction and personal growth derived from the tasks themselves, which does not address the root problem in the long-term.

Research Area: Self-Determination Theory (SDT)

The development of the adaptive scheduling algorithm is grounded in Deci's Self-Determination Theory (SDT) and Cognitive Evaluation Theory (CET), which highlights the importance of perceived autonomy and competence in order to build intrinsic motivation. Intrinsic motivation plays a dominant role in individual behaviour, but when external stimuli are controlling and pressure individuals to act a certain way, internal, self-initiated behaviour is diminished. Therefore, to achieve optimal function and growth, people need psychological liberty and independence over their situation and regular reinforcement of positive feedback to feel competent.

By granting users substantial control over their schedules, delivering continuous, real-time feedback on their task progression, and constantly readjusting the model based upon new, streaming data, the heuristics algorithm thereby augments one's perceived autonomy and competence. Furthermore, the "signalling" in the system integrates James Clear's "cue-craving-reward" framework for habit-building, where task input prompts and completion reminders serve as cues, progress visualization triggers cravings for task completion, and the reinforcement mechanism provides rewards, thereby solidifying positive behavioral patterns. The algorithm adjusts priorities based on user progress, establishing a visual and consistent feedback loop while still necessitating discipline in maintaining inputs. Ultimately, the goal is to provide positive reinforcement by conditioning individuals to take ownership of their own schedule, without becoming too overwhelmed when slipping behind.

Algorithm Layout

Heuristic algorithms employ practical shortcuts or pre-set logical patterns, often called "rules of thumb", to produce solutions prioritizing speed and efficiency to achieve immediate goals. They are effectively employed within the context of flow-shop scheduling problems, where a heuristic preference relation is developed and used as the basis for both job insertion and distribution to build up the complete schedule. For example, in a general flow-shop situation with the m-machine, n-job sequencing problem, where all jobs must pass through the machines in the same order, most heuristic algorithms propose that jobs with higher total processing time should be given higher priority than jobs with less processing time. But while industries often deal with fixed and repetitive tasks, personal scheduling, such as student study plans, incorporates a wide range of activities with varying degrees of cognitive demand for a wide range of individuals with different learning styles and preferences. Therefore, developing heuristic models that integrate psychometric assessments can enhance its predictability and adaptability to individual needs.

Algorithm Data Inputs

The algorithm will first "profile" the user, measuring their initial intrinsic motivation, task preferences and general study habits, deriving evaluation techniques from proven indexes like the Intrinsic Motivational Inventory (IMI). This data is sufficient to generate a basic regression model that outputs an Estimated Completion Time (ECT) for various tasks. Now, the user can input the various tasks they have, entering relevant information like due date, type of task and perceived level of difficulty for the task, which creates a task difficulty variable in the regression model. Based on task difficulty and the  study profile, the model outputs an ECT for each task. The algorithm now has to align the tasks and visually distribute them in a Gantt chart. As seen in the flow charts, there are two sequential considerations for distribution: task difficulty and task urgency (based upon the Eisenhower matrix of task prioritization), and a First-In-First-Out (FIFO) mechanism is also implemented, which is the basis of the heuristic preference relation.

Preliminary Data Collection and Expected Patterns

Data collection is largely in a preliminary stage for this project, but there have been several patterns that can be observed and predicted. Currently, the Task ECT prediction model has consistently placed a higher weight on the task difficulty variable over user motivation; a case that is corroborated by studies that have shown that task difficulty, which encompasses both perceived and actual effort required, is a more direct factor in determining how long it takes to complete a task. Results from Dominick Scassera's 2016 experiment showed that perceived difficulty had a significant linear effect on task performance, with higher perceived difficulty correlating with lower performance. 

The task ECT Prediction model should also presumably generate a consistent range of ECT multiplier values for similar types of study profiles. While not enough data has been collected to sufficiently solidify a range of task ECT multiplier values for various study profiles, existing psychological theory suggests that there will be a pattern in which lesser motivated users will have a larger performance gap with averagely motivated users than highly motivated users do with averagely motivated users. For example, individuals with lower motivation (Type S) may require 50% more time to complete tasks compared to those with average motivation (Type M), aligning with a 1.5 multiplier. Conversely, highly motivated individuals (Type D) can complete tasks about 25% faster, supporting the 0.75 multiplier. This is especially true when task difficulty increases, as studies by Locke and Latham demonstrate that high and specific goals increase performance and motivation, but the effect of goal difficulty on performance is more pronounced for those with lower initial performance levels (2012). As such, there is a higher "learning curve", which means that the Task ECT prediction model is expected to generate higher multiplier values. Inputting such "rules of logic" into the regression model will improve its accuracy, and may generate insights into the nature of motivation, effectively quantifying it.

Future Outlook

The future outlook for the adaptive scheduling algorithm involves its evolution into a user-friendly application incorporating gamification elements to enhance engagement. If such an app was integrated within educational platforms and workspaces, it could incorporate an additional source of data: streaming data based on application analytics regarding app usage, live task progress, and other interactive features that also boost direct user engagement.

There is also a goal to refine the model through extensive data collection. By integrating more heuristics derived from extensive data points, the algorithm will become more sophisticated and able to create context-aware, intuitive scheduling decisions. In the future, such continuous research and user feedback could possibly further optimize the scheduling process by minimizing the need for manual user inputs. Leveraging sophisticated AI and machine learning techniques, such as genetic algorithms and advanced optimization methods, will refine the algorithm's precision and adaptability, especially the "adjustable" feature. With such features integrated, this model holds much promise as a tool to maximize efficiency while minimizing anxiety and procrastination, boosting intrinsic motivation.