Design

Input hierarchy

Updated: Mar 13, 2026

The Horizon OS platform supports a wide range of input modalities, including spatial input sources like controllers and specialized devices that offer improved precision, throughput, and convenience. The input hierarchy defines how the system prioritizes these modalities and manages fallbacks, so users can transition gracefully between inputs depending on their context.

Input tiers

Horizon OS organizes input modalities into three tiers based on form factor and task coverage.

Tier 1 comprises input methods inherent to the device, catering to the broadest audience.

Tier 2 and Tier 3 offer enhanced precision or specialized input through software enhancements or accessories.

On devices that support eye tracking, gaze provides an additional targeting method. See Fallback behavior for how the system integrates gaze when available.

Tier	Description	Modalities
1. Core	The most natural and high-priority modalities available	Controllers, Hands
2. Core and specialized	Functional alternatives that enhance precision, throughput, or convenience	Gamepad, Keyboard, Mouse, Stylus
3. Specialized	Supplementary or system-level inputs used for specific tasks	Voice

Note: On devices that support eye tracking, gaze can serve as an additional targeting method for far-field interactions. When eye tracking is enabled and controllers are not in hand, the system can use gaze for targeting paired with a hand pinch gesture for selection. See Fallback behavior for details.

Tier 1: Core interaction model

Input methods that serve the largest content and task types and deliver a core interaction model that meets most users’ needs.

Controllers
Controllers are a crucial input modality on the Meta Quest platform, enabling sophisticated spatial input that allows users to perform tasks requiring precision and specific functions. Thanks to its exceptional versatility and precision, the controller is an ideal input device for a broad spectrum of applications.

Hands
Hand tracking enables the use of hands as an input method. Using hands as an input method delivers a new sense of presence, enhances social engagement, and delivers more intuitive and natural interactions.

Tier 2: Between core and specialized

Input devices or modalities that some users may use to increase precision, throughput, convenience, and more capabilities for under-performing input in Tier 1.

Gamepad
Gamepad input is specialized for gaming experiences with various analog and digital buttons combined with D-pad and thumbsticks. Gamepads offer precise control over movements and actions, making them well-suited for games and applications requiring accuracy.

Keyboard
Meta Quest devices support physical keyboards through a Bluetooth connection. This input modality allows users to input text quickly and efficiently, leveraging the familiarity and muscle memory of typing on a keyboard.

Mouse
Mouse input enables precise control over interactions, making it ideal for tasks like selecting small objects or navigating complex menus. Users are accustomed to using mice in traditional computing, which makes it an intuitive input method.

Stylus
Stylus input allows the user to draw and sketch on both 2D surfaces and in 3D space. Stylus input offers high precision and accuracy, making it ideal for tasks like writing, drawing, and selecting small objects.

Tier 3: Specialized input

Supplementary or system-level inputs used for specific tasks or for smaller segments of users.

Voice
Voice input enables users to interact with devices and applications using spoken language. This modality is especially useful for hands-free interaction, accessibility, and situations where manual input is impractical.

Fallback behavior

When a preferred input method is unavailable or becomes obscured, the system uses fallback logic to maintain functionality. The following transitions happen automatically.

Targeting fallback

The system defaults to Controller Ray or Hand Ray for far-field targeting. On devices with eye tracking, if controllers are not in hand and eye tracking quality is sufficient, the system can use gaze for targeting instead. If no hand or controller tracking is available, the Head ray (center of FOV) becomes the fallback aimer.

Default priority: Controller/Hand Ray > Head ray

With eye tracking enabled (controllers not in hand): Gaze > Hand Ray > Head ray

For more information, see Ray casting.

Selection fallback

The primary selection method for hands is the Pinch gesture, and for controllers, it is the Trigger button. If these cannot be detected, the system falls back to the physical Volume buttons on the headset as a universal “click” signal.

Priority: Hands / Controllers > Volume buttons

For more information, see Touch and Grab.

Distance transitions

The hierarchy shifts based on proximity. Ray casting is enabled when the user’s hand or controller is far from an interactable. If the hand or controller gets closer, the ray is automatically turned off, and direct interaction is prioritized.

For more information, see Multimodality and Input mappings.

Input transition

Horizon OS is designed to provide a seamless handoff between modalities. It prioritizes active hardware devices over hand tracking to reflect user intent. When a user picks up a controller or mouse, the system automatically transitions to device-specific inputs.

For more information, see Multimodality.

Hand gesture hierarchy

Beyond the modality tiers above, hand tracking has its own internal gesture ranking.

Gestures are ordered by reliability, whether they are CV tracked or ML trained, and importance in order for the user to interact/navigate with the system. Since users can only comfortably memorize a few gestures, maintaining consistency across the system and applications is essential.

Tier 1

Pinch	ML trained. High reliability. Required hand gesture. The universal selection gesture where the thumb and index finger touch. It is essential for indirect selection with a targeting method, and for direct selection to hold or pick up interactables, mimicking real-world small object interaction. For more information, see Ray casting and Grab.
Palm up pinch	ML trained. High reliability. Required hand gesture. With the palm facing up, the thumb and index finger meet. This is used with the dominant hand to call the system menu and with the non-dominant hand for the app or context menu.

Tier 2

Microgesture	ML trained. High reliability. Secondary hand gesture. Small, thumb-based gestures performed on the side of the index finger. This gesture expands hand-based inputs, enabling users to perform low-effort navigation and shortcuts for high-frequency interactions. For more information, see Microgestures.
Palm grab	ML trained. High reliability. Secondary hand gesture. Gestures that involve multiple fingers closing to grab an interactable. While pinch is intuitively used for smaller items, users perform a palm grab gesture when the interactable is bigger. For more information, see Grab.

Tier 3

Static pose	No ML. Variable reliability. Custom hand gesture. Hand poses—such as a "Peace" sign or a "Thumbs-up"—that are specific to an app and are not part of the core Horizon OS machine-learning-trained set. The responsibility for training and reminding the user of these gestures lies entirely with the application.
Expressive gestures	No ML. Variable reliability. Custom hand gesture. Active hand poses in motion—such as waving—that are specific to an app and are not part of the core Horizon OS machine-learning-trained set. The responsibility for training and reminding the user of these gestures lies entirely with the application.