Semantic Pointing:Improving Target Acquisition withControl-Display Ratio AdaptationRenaud BlanchLRI & INRIA Futurs∗Universit´e Paris-SudOrsay, [email protected] GuiardMouvement et PerceptionCNRS & Universit´e dela M´editerran´eeMarseille, [email protected] Beaudouin-LafonLRI & INRIA Futurs∗Universit´e Paris-SudOrsay, [email protected] introduce semantic pointing, a novel interaction tech-nique that improves target acquisition in graphical user inter-faces (GUIs). Semantic pointing uses two independent sizesfor each potential target presented to the user: one size in mo-tor space adapted to its importance for the manipulation, andone size in visual space adapted to the amount of informationit conveys. This decoupling between visual and motor size isachieved by changing the control-to-display ratio accordingto cursor distance to nearby targets. We present a controlledexperiment supporting our hypothesis that the performanceof semantic pointing is given by Fitts’ index of difficulty inmotor rather than visual space. We apply semantic pointingto the redesign of traditional GUI widgets by taking advan-tage of the independent manipulation of motor and visualwidget sizes.Categories and Subject Descriptors: H.5.2 [InformationInterfaces and Presentation (e.g., HCI)]: User Interfaces –Graphical user interfaces, Input devices and strategies, In-teraction styles, Theory and methodsKeywords: Control-display ratio, Fitts’ law, graphical userinterface, pointing, semantic pointingINTRODUCTIONPointing is a fundamental task in graphical user interfaces(GUIs). To help manage the growing complexity of soft-ware, such as the increasing number of toolbars and menucommands, the HCI literature has introduced new interactiontechniques that attempt to reduce pointing time. This paperexplores the idea of assigning two separate sizes for objectsin the interface: a visual size for display, and a motor sizereflecting the importance of the object for interaction. We∗projet InSitu – Pˆole Commun de Recherche en Informatique duplateau de saclay – CNRS,´Ecole Polytechnique, INRIA, Universit´eParis-Sud.Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copiesare not made or distributed for profit or commercial advantage and thatcopies bear this notice and the full citation on the first page. To copyotherwise, or republish, to post on servers or to redistribute to lists,requires prior specific permission and/or a fee.CHI 2004, April 24-29, 2004, Vienna, Austria.Copyright 2003 ACM 1-58113-702-8/04/0004. . . $5.00.hypothesize that task difficulty depends on the motor, notvisual size of objects, and control the motor size by adapt-ing the control-display (C-D) ratio. We call this techniquesemantic pointing, since motor sizes are used to reflect thelocal semantics of the screen.Fitts’ law [6] is widely used to design and evaluate interac-tion techniques and input devices [16]. It links the movementtime (MT) to acquire a target to the task’s index of difficulty(ID). ID is the logarithm of the ratio between target distance(D) and target width (W). MT is a linear function of ID char-acterizing the system. The implications of Fitts’ law havebeen used in several techniques to facilitate pointing tasksby enlarging the target or by reducing its distance [4, 1, 18,5, 24, 9].C-D ratio(m/pixel)speedspeedC-D ratio(m/pixel)(a) (b)Figure 1: C-D ratio as a function of mouse speed(a) constant C-D ratio (b) mouse accelerationControl-display ratio adaptation [13, 23, 5] is another ap-proach for facilitating target acquisition. This technique im-proves pointing performance but has not been analyzed interms of Fitts’ law, and its possible use in real GUIs has notbeen fully explored. The C-D ratio [17] is a coefficient thatmaps the physical displacement of the pointing device to theresulting on-screen cursor movement in a system where thereis an indirection between the pointing device and the display(typically with a mouse). The C-D ratio defines the distancethe mouse has to cover in the physical world (dx in meter)to move the cursor on the screen by a given distance (dX inpixel)1. The C-D ratio is dx/dX. A typical C-D ratio adapta-tion is the so-called mouse ‘acceleration’. The cursor movesby a larger distance when the mouse covers a given ampli-1We use the following conventions: capital letters denote quanti-ties (e.g. distances) concerning the screen, while lower case lettersdenote those concerning the physical world. For distances, we usetwo different units (pixel and meter respectively) to disambiguateratios that would otherwise be dimensionless.CHI 2004 ׀ Paper 24-29 April ׀ Vienna, Austria Volume 6, Number 1 519tude more quickly (Figure 1), capturing an intention: whenusers move the physical device quickly, they typically wishto go further, so the cursor can be displaced even faster tocover the distance more quickly.After reviewing previous work on facilitating target acqui-sition, we describe semantic pointing and predict its effecton pointing performance in terms of Fitts’ law. We then de-scribe a controlled experiment that tests our predictions. Fi-nally, we illustrate potential applications of semantic point-ing to GUI design.RELATED WORKGrowing Target and Shrinking DistanceWith respect to Fitts’ law, there are two simple ways to re-duce the difficulty of a pointing task: enlarging the target ormoving it closer to the cursor. Both have been explored inseveral ways. A widely-used direct application of this prin-ciple is contextual pop-up menus. Such menus are displayedat the cursor location so that distances to the items are mini-mal. Pie menus [4] are even more radical: the distance fromeach menu item to the cursor is constant and very small. Thedistance can also be reduced by moving the potential targetsof a directed movement towards the cursor, as in the drag-and-pop technique [1].Another approach consists in modifying target size when thecursor is close enough. This can be achieved by magnifyingthe target [18], or by adding a ‘bubble’ around it [5]. Eval-uations and comparisons with other techniques [23, 5] showthat target resizing facilitates pointing even if the expansionis late [18] and unpredictable [24]. The problem in apply-ing such techniques to real GUIs is that in order to expand atarget surrounded by other possible