Extensive research has examined the ability to process multiple tasks in parallel, with recent studies highlighting systematic individual preferences for serial versus parallel processing modes. To determine whether these modes reflect traits or states, we investigated their dependence on task environment predictability and volatility using the task switching with preview paradigm (TSWP). In classical TSWP, participants alternate predictably between two displayed tasks, with the preview of the currently irrelevant task allowing response preparation. Key findings revealed that some individuals processed the preview in parallel (overlapping processing) to achieve switch benefits, whereas others opted for serial processing. We investigated whether these processing modes reflect traits or states by manipulating task predictability. In Experiment 1 (N = 60), we introduced blocks with variable task sequences, in which tasks alternated randomly every second to fourth trial. Results showed a marked shift toward more serial processing in variable relative to fixed sequence blocks, suggesting reduced use of the preview under unpredictable conditions. To determine whether predictability or context volatility caused these results, Experiment 2 (N = 60) added a countdown to the variable sequence blocks, making task switches predictable while maintaining context volatility. Unlike Experiment 1, the data revealed only a slight reduction in the use of the preview across varying sequence lengths, indicating that parallel processing is influenced by task predictability rather than context volatility. These findings suggest that individual differences in processing modes are modulated by state-like factors such as task predictability. Implications for understanding flexibility in individual multitasking preferences are discussed.