Trade-offs in the Design of Multimodal Interaction for Older Adults

Abstract

This position paper presents key aspects and trade-offs that designers, ICT and HCD practitioners might encounter when designing multimodal interaction for a specific target group: older adults. The paper draws together literature on multimodal interaction and assistive technology and presents a set of specific findings for older users. Building on these findings, we describe a number of challenges that should be taken into account when designing multimodal technology for this target group. Further reflections highlight the design trade-offs that such considerations might bring, presenting an overview of the design choices and their potential effects.

Keywords

• Multimodality
• older adults
• design trade-offs

Introduction

Multimodal interfaces seek to combine multiple sensory input and output channels in similar ways as in natural interaction. This similarity has led to the expectation that multimodality in human-computer interaction can provide a more natural, robust and flexible form of interaction with respect to more traditional input modalities such as mouse and keyboard. [1, 2]. In this respect, multimodal human–computer interaction has sought to provide not only more powerful and compelling interactive experiences, but also more accessible interfaces to technological devices. Moreover, following the principle of “design for all” and “inclusive design”, multimodal technology has been proposed as a possible solution that allows users to use the modality matching their preferences and/or needs, thus making the interaction more flexible [3]. However, despite these potential advantages of multimodal interfaces, the literature reports significant disadvantages. For example, different modalities may interfere with each other and a synchronization problem might arise.

Additionally, combining and coordinating more than one modality might also require more effort for the users [4, 5] and a higher cognitive load [4]. Current research provides findings supporting both assumptions reporting advantages [1, 2] as well as disadvantages [4, 5]. This position paper has the goal to further advance the discussion on this topic by presenting design trade-offs in multimodal technology when designing technology for older adults, a large and heterogeneous user population.

Multimodal Interaction for Older Users

Multimodal interfaces have been considered to improve accessibility for a number of users and usage contexts [3], including the diverse needs of older users [ 6, 7 ]. Multimodal systems can integrate a wider range of modalities (such as speech, writing, gaze, touch or mid-air gestures) and better accommodate users’ preferences with respect to unimodal interfaces. Furthermore, multimodal interfaces can be more userfriendly for people who have little or no experience with common computer devices, since they offer the possibility to use multiple interaction channels without relying on a single source of input [3]. However, other studies point out that multimodality must be carefully introduced since it might require more cognitive effort to coordinate different input modalities (especially when more than two modalities are involved) and additional physical demand [4]. This may become particularly relevant when considering the cognitive and physical characteristics of older users [ 8 ].

Numerous examples of multimodal technology for older adults can be found in the research and on the market. For instance, social robots or telepresence technology are two representative examples of multimodal systems believed to support older users [ 7 ]. Before listing the design trade-offs that these or similar interaction modalities might bring in the technology, we summarize a list of design considerations. 3

Design Challenges and Recommendations

Different recommendations for the design of multimodal interfaces for older adults have been discussed in the literature (see for example, [ 4, 7, 9 ]) and are summarized below.

Diverse abilities. Multimodal systems provide users with the choice of using the most efficient interaction modality among those proposed by the system. Moreover, users can switch to another interaction modality after a recognition error has occurred in the previous one [2]. However, this requires that the user knows which is the best modality for her/him, or at least “intuitively” uses the best set of multimodal inputs. Personalization. One-solution-fits-all models are inadequate as they do not consider individual characteristics. The interaction and the interface should be made adaptable and personalized to user preferences and device characteristics. Users should be able to customize the multimodal channels they want to use for a given task in an application. Interaction patterns. There are large individual differences in users’ multimodal interaction patterns [ 10 ]: some individuals tend to integrate different modalities in a simultaneous manner and overlap them temporally (simultaneous integrators), whereas others tend to finish one mode before beginning the second (sequential integrators). Older adults demonstrate either a simultaneous or sequential dominant integration pattern [ 10, 11 ], however designers should be aware of such differences and multimodal interfaces should accommodate individual interaction patterns. Independence. Multimodal interfaces should empower older users to independently interact with the technology, even when there is a specific impairment (for example hearing loss or reduced sight). Multimodal interfaces can also contribute to seniors’ perceived independence [ 7 ], if they can empower the user to function independently. Technology reliability. Users should be able to rely on the multimodal technology, especially in the case of assistive technology. For this reason, multimodal processing should be accurate and robust. However, since recognition algorithms are mainly trained on data from non-older population, this might pose limitations on the performance (e.g., due to vocal features characteristics of older adults for speech recognition [ 12 ], or slower gesture speed for gesture recognition).

Privacy and context of use. Context of use should be carefully considered when designing multimodal technology [ 13, 14 ]: older people have privacy and social acceptability concerns about using some modalities in public spaces (as in the case of speech commands or mid-air gestures [ 14 ]). However, one of the advantages of multimodal interaction is the possibility of using one modality rather than the other according to the specific context (e.g., gestures instead of speech commands in noisy environments). 4

Design Trade-offs

When considering the recommendations mentioned above, practitioners and designers might experience a number of design trade-offs, which are situations that involve losing one quality or aspect of the design in return for gaining another quality or aspect. Table 1 presents a summary of design challenges and design trade-offs.

Complexity vs Simplicity. Providing the user with the possibility to interact with more than one modality might increase the interaction complexity. For example, it has been showed that the oldest of older adults (80+) found some modalities or combination of modalities too complex to use when using a multimodal tablet-based application [ 13 ]. This trade-off might also affect the system usability: a technology that supports many different modalities might increase in complexity and thus be less usable. Personalization vs Customization. Multimodal interaction can be tailored to the specific preferences or needs of the user. This process might end up in an overpersonalization of the interaction, making it difficult to the user to discover or experiment with alternative interaction modalities. There is indeed a trade-off between personalization, where the system personalizes the interaction to the user, and customization, where the user is in control of the customization process. The latter allows the users to control the interaction, assuming that they know how and what feature to control. The former gives control to the system without requiring an effort from the user, but it heavily relies on system reliability and performance. Independency vs Assistance. The cognitive effort required to older users for the personalization and customization of system interaction may be avoided by allowing other users to take care of the process. For instance, a multimodal technology could be designed to be personalized by caregivers or therapists even though the final user is an older adult (assistance). However, delegating actions to these users might further increase the caregiver load and could be perceived as an additional demand or burden. This might also decrease older adults independent use of the technology. Automation vs Control. Multimodal sensors can be used as background controls, to which the interface automatically adapts without any intentional and direct user engagement. In this sense, a proactive system might come forward with suggestions, or automatic responses, based on the sensed context and without engaging the user (automation). On the other hand, a reactive system requires the user to initiate action (control), which implies direct attention and control on the activity. 5

Conclusions

This position paper has briefly presented some of the challenges and trade-offs that designers, ICT and HCD practitioners might encounter when designing multimodal interfaces. Even though the analysis focuses on older adults as target user group, most of these challenges hold also for the wider user population.

On one side, this paper presents a reflection on how to identify the design trade-offs for multimodal interaction for older adults. On the other side, further discussion is needed to provide practitioners with methods and processes for dealing with such tradeoffs. In this direction, participatory design, co-design and value-centered design approaches [ 15 ] can help designers to balance different (and sometime competing) design choices. We think that further discussion on this topic can emerge from the discussion during the workshop.

Acknowledgments

This work is supported by the EU HORIZON project ECOMODE - Event-Driven Compressive Vision for Multimodal Interaction with Mobile Devices (http://www.ecomode-project.eu/), under Grant Agreement 644096.

References

Oviatt , S. , Cohen, P. : Perceptual User Interfaces: Multimodal Interfaces That Process What Comes Naturally . Commun. ACM . 43 , 45 - 53 ( 2000 ).

Turk , M. : Multimodal interaction: A review . Pattern Recognition Letters . 36 , 189 - 195 ( 2014 ).

Obrenovic , Z. , Abascal , J. , Starcevic , D. : Universal Accessibility As a Multimodal Design Issue . Commun. ACM . 50 , 83 - 88 ( 2007 ).

Naumann , A.B. , Wechsung , I. , Hurtienne , J.: Multimodal Interaction: A Suitable Strategy for Including Older Users? Interact . Comput. 22 , 465 - 474 ( 2010 ).

Wechsung , I. , Naumann , A.B. : Evaluation Methods for Multimodal Systems: A Comparison of Standardized Usability Questionnaires . In: André, E. , Dybkjaer , L. , Minker , W. , Neumann , H. , Pieraccini , R. , and Weber , M. (eds.) Perception in Multimodal Dialogue Systems . pp. 276 - 284 . Springer Berlin Heidelberg, Berlin, Heidelberg ( 2008 ).

Himmelsbach , J. , Garschall , M. , Egger , S. , Steffek , S. , Tscheligi , M. : Enabling Accessibility Through Multimodality?: Interaction Modality Choices of Older Adults . In: Proceedings of the 14th International Conference on Mobile and Ubiquitous Multimedia . pp. 195 - 199 . ACM, New York, NY, USA ( 2015 ).

7. Munteanu , C. , Salah , A.A. : Multimodal technologies for seniors: challenges and opportunities . In: Oviatt, S. , Schuller , B. , Cohen , P.R. , Sonntag , D. , Potamianos , G. , and Krüger , A . (eds.) The Handbook of Multimodal-Multisensor Interfaces . pp. 319 - 362 . Association for Computing Machinery and Morgan & Claypool, New York, NY, USA ( 2017 ).

8. Fisk , A.D. , Rogers , W.A. , Charness , N. , Czaja , S.J. , Sharit , J.: Designing for Older Adults: Principles and Creative Human Factors Approaches, Second Edition . CRC Press ( 2009 ).

9. McGee-Lennon , M.R. , Wolters , M.K. , Brewster , S.: User-centred Multimodal Reminders for Assistive Living . In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems . pp. 2105 - 2114 . ACM, New York, NY, USA ( 2011 ).

10. Xiao , B. , Lunsford , R. , Coulston , R. , Wesson , M. , Oviatt , S. : Modeling Multimodal Integration Patterns and Performance in Seniors: Toward Adaptive Processing of Individual Differences . In: Proceedings of the 5th International Conference on Multimodal Interfaces . pp. 265 - 272 . ACM, New York, NY, USA ( 2003 ).

11. Schiavo , G. , Mich , O. , Ferron , M. , Mana , N.: Mobile Multimodal Interaction for Older and Younger Users: Exploring Differences and Similarities . In: Proceedings of the 16th International Conference on Mobile and Ubiquitous Multimedia . pp. 407 - 414 . ACM, New York, NY, USA ( 2017 ).

12. Vacher , M. , Portet , F. , Rossato , S. , Aman , F. , Golanski , C. , Dugheanu , R.: Speechbased interaction in an AAL-context. 1-7 ( 2012 ).

13. Neves , B.B. , Franz , R.L. , Munteanu , C. , Baecker , R. , Ngo , M. : “ My Hand Doesn'T Listen to Me! ”: Adoption and Evaluation of a Communication Technology for the 'Oldest Old .' In: Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems . pp. 1593 - 1602 . ACM, New York, NY, USA ( 2015 ).

14. Rico , J. , Brewster , S. : Gesture and Voice Prototyping for Early Evaluations of Social Acceptability in Multimodal Interfaces . In: International Conference on Multimodal Interfaces and the Workshop on Machine Learning for Multimodal Interaction . pp. 16 : 1 - 16 : 9 . ACM, New York, NY, USA ( 2010 ).

15. Fischer , G. : Exploring design trade-offs for quality of life in human-centered design . interactions . 25 , 26 - 33 ( 2017 ).