The Re-Iterating Maze helps inmates at Alcatraz develop meditative practices through rehabilitation, using their sentence to make a more peaceful existence possible. Our concept allows prisoners a mindful form of exercise, allowing them to use their time outdoors exploring multidimensional pathways.
Mindful walking encourages the brain to focus on the activity of traversing a determined path. By minimizing choices and instead focusing on the journey, stress is reduced and the traveler may learn more about how their mind works. While this traditionally occurs in natural environments, it can be performed anywhere.
Early explorations of our concept involved building organic forms draped over the site to create stimulating pathways. Through this, we wanted to establish varying design languages for individuals to be able to express themselves and not be bound by the grey dreariness that haunts prison cells.
The computational methods used to generate interactive walking routes are perlin noise generation, voxelization, surface mapping, and depth first search.
Figure 1: Early Form Exploration
Figure 2: Early Form Exploration
Figure 3: Early Form Exploration
We settled on the idea of a maze-like path generation, where the person walking through is exposed to a new iteration of the maze each time they experience it, with different inputs. These are then mapped onto varied surfaces to create a more organic experience, similar to hiking through hills and emulating the calmness that accompanies natural excursions.
We made a custom python script that generates bespoke paths without any branching. Slider inputs allow the complexity and hilliness of the maze to be increased.
Figure 4: Unique Maze Generation with Same Complexity
Figure 5: Iterations with Increasing Complexity
Figure 6: Mapping Same Maze on Iterations of Hilly Suface
Additional walking paths are generated to complement the iterating maze, using multiple input parameters to determine which routes can be traversed on the island.
Perlin noise generation was placed systematically around a curve representing the open area outside the prison. After multiple iterations, we selected the iteration that best suits the topology and the entrances from the prison complex. An area in the center of these paths is kept available for the maze, which is intentionally placed behind less challenging modes of walking for a gradual inclusion of complexity.
Figure 7: Form Finding of Additional Walking Paths
The table below illustrates the logic followed to design and create this parametric and evolving structure:
Figure 9: Logical Flowchart
The maze follows the chart shown below, where the y axis shows an increase in the complexity, and the x axis the amplitude of the hill the maze is placed upon. The top-left maze is the simplest, the bottom-right maze is most complex.
Figure 10: Maze Iteration Table
The voxelized pathways were given varying seeds for voxel size, grid size, scale, octaves, and speed.
Figure 11: Walking Path Iterations
Figure 12: Selected Walking Path Iteration (Iteration 8)
We chose Iteration 8 for its relationship to the site and the optimal amount of voxel change. Iteration 8 best allows prisoners to make their way around the island.
Figure 13: Form without Site Context
Figure 14: Plan View of Form and Site Context
Figure 15: Human Perspective View of Form
Figure 16: Aerial View of Form and Site Context
Figure 17: Section of Form with Site Context
Figure 18: Section of Form with Site Context
The Reiterating Maze is a project of IAAC, Institute for Advanced Architecture of
Catalonia developed at MaCAD 2025 by students: Mahnoor Fatima and Scott Lebow, and faculty: James McBennett, Hesham Shawqy, and Eva Papaspyrou.
