What's so hard about hierarchical control? Pinpointing processing constraints within cue-based and serial-order...

Area of Science:

• Cognitive Science
• Neuroscience
• Human-Computer Interaction

Background:

• Most complex tasks involve hierarchical structures, where decisions depend on preceding higher-level choices.
• The precise impact of increasing hierarchical levels on task difficulty remains unclear, with competing theories on cost localization (local updating vs. global integration).
• Task navigation can be cue-based (environmental prompts) or serial-order based (sequential plans), potentially influencing cognitive demands.

Purpose of the Study:

• To investigate how the number of hierarchical levels affects task performance.
• To differentiate between local updating costs and global integration/maintenance costs in hierarchical tasks.
• To examine the influence of control format (serial-order vs. cue-based) on hierarchical task costs.

Main Methods:

• Two experiments were conducted assessing performance across varying hierarchical levels and control formats (serial-order and cue-based).
• Computational modeling was employed to compare different theoretical accounts of cognitive costs.
• Performance metrics were analyzed as a function of hierarchical depth and control strategy.

Main Results:

• In the serial-order format, performance costs scaled with increasing hierarchical levels, consistent with global maintenance and integration demands.
• In the cue-based format, costs were observed primarily during context updating and with the inclusion of one additional relevant level.
• Beyond the immediately relevant higher level, additional levels in cue-based tasks did not significantly increase costs.

Conclusions:

• The cognitive costs associated with hierarchical task structures depend significantly on the control mechanism employed.
• Serial-order control necessitates integrating information across all hierarchical levels, leading to global costs.
• Cue-based control involves checking the immediate higher-level context, resulting in localized costs at updating points and for the next relevant level.