Olfactory Display Joseph Jofish Kaye Cornell Information Science

Olfactory Display Joseph ‘Jofish’ Kaye Cornell Information Science Culturally Embedded Computing UIST'04: Invited Survey 26 October 2004

Contents 0. Why should we care? 1. How smell works. 2. Typologies: Why smell is hard. (3 a)Technologies for Computerized Scent Output (3 b) Uses for Computerized Scent Output 3. Olfactory Display: History, Academia, Industry 4. Symbolic Olfactory Display: Theory 5. Conclusions: A Future for Computerized Scent Output

0. Why should smell matter to UI? 1. Because it makes us question our assumptions about how and what we can and should communicate information through a user interface 2. Because it's powerful and evocative sense, strongly tied into experience. 3. Because arguably the inherently ambient interface.

1. How We Smell Olfactory epithelium contains ~1 k kinds of receptors, each from a single gene. (Buck & Axel (1991) A Novel Multigene Family May Encode Odorant Receptors: A Molecular Basis for Odor Recognition. Cell (65)). 2004 Nobel Prize for Physiology or Medicine!

• Vibrational (Dyson 1938, Wright 1954) So, how do those smell • Lock & Key receptors (Amoore 1963) work? 3 • Electron theories: tunneling (Turin 1996) ~

Turin (1996) • Correlations between smell and molecular structure but not shape (bitter almond character of 40+ molecules including simple HCN, complicated benzonitrile) • Replace hydrogen in acetephenone with deuterium (heavy hydrogen): fruitier, more almond • Proposes inelastic electron tunneling as a mechanism • Smell, like colour vision and hearing, is a spectral sense Turin, L. A spectroscopic mechanism for primary olfactory reception. Chemical Senses 21(6) 1996.

Odor Quantity: We have ≤ 2 bits of precision Accuracy ratings for 100% to 0% rose concentrations: near chance Engen, Trygg and Pfaffman, Carl. (1959) Absolute judgements of odor intensity. Journal of Experimental Psychology, 58(1): 23 -26.

Odor ‘Quality’ How Many? • Engen & Pfaffman 1960: 4 bits. • Everyone else, later: Hundreds. • Why the difference? Naming isn’t remembering: • ‘Tip of the nose’ phenomena: Labeling is everything: • “Fishy-goaty-oily” vs. leather • Complex smells are easier: coffee vs. l-carvone

Varience in individuals and population Individuals vary significantly in their abilities to smell different odors: specific anosmia An individual’s variance over time is as high as the varience in the general population. (Stevens et. al. 1988)

Adaptation and Mixtures People adapt to ambient smells in under a minute (Cain 1974) A strong smell smelt half an hour ago will effect how you perceive smells now. (Hulshoff Pol et. al 1998) Mixing smells has unpredictable results on the ability to perceive each.

So… Conclusion #1: Using the strength of a scent to convey information is unreliable. You must use different aromas.

Section Two: Typologies Why is smell hard? How can we organize smells?

Why is smell hard? What if we could only use referents to refer to colour? "fire-engine colour" "grass colour"

Why is smell hard? …which is what we do with smell. "mint" ("smells like mint smells like. ") "peppery", "blackberry", etc.

Why is smell hard? Exceptions: trigeminal sensations: "spicy hot" "minty cold" taste-related (tricky): "sweet", "metallic"

Why is smell hard? No clear system for organizing. All such systems are strongly culturally embedded. Some examples:

Typologies I First 'modern' system: generally attributed to Linnaeus (1756); actually by Andreas Wahlin, U Gothenberg • Aromaticos aromatic • Fragrantes fragrant • Ambrocacos ambrosial (musky) • Alliaceos alliaceous (garlicy) • Hircinos hircine (goaty) • Tetros repulsive • Naufeofos nauseous Zweibakker (1895) 9 categories, above + "ethereal (fruits)" + "burnt (Empyreumatic)" Thanks to Janet Vertesi for Latin translation

Typologies II: Sua Indians, Brazil bland-smelling: adult men small mammals, birds most fish innocuous plants pungent-smelling: old men, women large mammals, macaw some amphibians medicinal plants strong-smelling: adult women, children, birds, carnivorous mammals harmful plants Classen, Howes and Synnott (1994) Aroma: A Cultural History. Routledge

Typologies: Conclusions “There have been numerous attempts at the classification of odors, all of which have been empirical and useless and most of which have been childish and absurd. ” E Parry. The Class. of Odors Perf. Ess. Oil Rec. 7 129 -132 1916 Sagarin, E. On the inherent invalidity of all current systems of odor classification. Journal of the Society of Cosmetic Chemists. p 25 -35 Vol. 2 No. 1 1950. There is no good general purpose Pantone-like system.

Technologies for Computerized Scent Output • Airbrushes (large volume, complicated, little control) • Solenoid & perfume bottle (good for prototyping; little quantity control) • Fan, and/or heat: scented wax, liquid, polymer (little control) • Inkjet (expensive; hard, lots of control. (HP, aromajet. com)) • Mechanized scratch-n-sniff (small volume, lots of control) • Add to existing HVAC (large volume, little control)

• Scent for its own sake Uses of • Computerized Scent Output – perfume counters, kiosks – functionary (aromatherapy, etc) Accompanying other media to augment an experience (movies, VR, websites, games, physical spaces…) • Symbolic olfactory display (warning signals, etc)

Section Three: Olfactory Display History Movies Patents Academia Industry

Movies Ambient: Aromarama 1958 Smell-o-Vision 1959 Scratch & Sniff: Polyester 1981 ENO, Ren & Stimpy, etc.

Sensorama Morton Heilig, 1962 Rheingold, H. Virtual Reality, Ch. 2. Summit Books, NY. Heilig, Morton. US Patents #3, 050, 870, #3, 469, 837, #3, 628, 829.

4. Academia

Virtual reality firefighter training: Di. Ve J. Cater (1993 -98)

R. Strong, W. Gaver Feather, Scent and Shaker: Supporting Simple Intimacy CSCW ‘ 96

Dr. Jenny Tillotson Senior Research Fellow, Central St. Martins School of Fashion & Design, Scentsory Design initiative. Tillotson, Jenny. (1997) Interactive Olfactory Surfaces. The Wellness Collection - A Science Fashion Sto Ph. D Thesis, School of Fashion & Textiles Research, Royal College of Art, London.

Interactive scent-emitting jewelry, senses physiological responses. Tillotson: Collaboration with nanotechnologists Current (Professor Manz, Inst. work: e. scent of Spectrochemistry & Applied Sciences)

Kaye (2001, 2004) in. Stink (12 bit presence & activity awareness) Dollars & Scents (2 bit obligatory ambient stock market display) Scent Reminder (n x 1 -bit) Honey I’m Home (1 bit minimal presence awareness) Kaye, J. Symbolic Olfactory Display. MS Thesis, MIT Media Lab. Kaye, J 'J'. Making Scents: Aromatic Output for HCI interactions jan+feb 2004

in. Stink

Dollars & Scents 1 -bit ambient, peripheral, NASDAQ stat

Scent Reminder

Honey, I’m Home

Directed Scent: Yasayuki Yanagida, ATR, Kyoto. Olfactory cannons for spotlight-like scent control Nose finding (it's between the eyes and down a bit) (video) Y. Yanagida, et al. Projection-Based Olfactory Display with Nose Tracking, Proceedings of IEEE VR 2004, pp. 43 -50, Chicago, U. S. A. , March 2004

Olfactory Cannons a brief note - Essentially sawtooth wave through chamber with small outlet port. - Can scale up to very large sizes (55 gallon drums, larger for crowd control) - Excellent description at http: //www. howtoons. org - Video Joost Bonsen & Saul Griffith. IGNobels 2003. www. howtoons. org

SIGGRAPH 2004: Scents of Space Usman Haque, J Pletts, L Turin Laminar airflow to provide selective space scenting (video)

Unconventional human computer SIGGRAPH interfaces tutorial. 2004, ACM VRST Steffi Beckhaus, Ernst Tutorials Kruijff

5. Industry

Patents: Many. 3 M, IBM, IIT, Motorola. All tech companies, not smell. Kaye, J. Computer Controlled Scent Output, Perfumer & Flavorist 28(6) 2004

Digiscents ~1998 -2001 $20 M venture capital 70 employees Some software No hardware Went bust, April 2001. Platt, Charles. You've Got Smell. Wired 7. 11 November 1999 Abramson, R. Sniff-Company Digiscents is a Scratch. The Register 11 April 2001.

Trisenx Production units 'next week' $369; educational market (olfactory alphabet) 20 smells; USB www. trisenx. com

Collaborations with France Telecom, Telewest Broadband Olfacom (UK), Kaori (Japan) 1 -bit (€ 100), 6 -bit (Michel (€ 700) scent output Pozzo) devices Probably doing the most in the field www. olfacom. com

Osmooze Single ($/€ 35), six bit ($/€ 329) dispensers. USB controlled. Collaboration with France Telecom www. osmooze. com

Dale. Air (Frank Knight) Scented spaces, deodorizing, museums, from £ 45. 97 Not fundamentally designed for computer control www. daleair. com Aggleton, J. P. , Waskett, L. (1999) The ability of odours to serve as statedependent cues for real-world memories: Can Viking smells aid the recall of Viking experiences? British Journal of Psychology 90: 1 -7.

aerome Desktop/kiosks for scent output Marketing emphasis (but collaborate, with, support Tillotson) www. aerome. com

Large systems for architectural Aromasys scenting (casinos, (Mark hotels, museums) Peltier) HVAC based www. aromasys. com

Scent. Air (Forrest Fleming) Commercial, marketing Scent. Pop ($200), Scent. Wave ($500), Scent. Blitz ($1000) + $25/cartridge www. scentair. com

P&G Scentstories Generates a new scent every 30 mins; 5 scents per disk. $35 machine + 1 disk; $5/disk. Febreeze branded; alternative to plug-in air freshener Why? Home fragrance $2. 7 B in 2002; projected $3. 5 B in 2006

4. Symbolic The use of controlled scent Olfactory output to display Display information. Kaye, J. Symbolic Olfactory Display. MS Thesis, MIT Media Lab. Kaye, J 'J'. Making Scents: Aromatic Output for HCI interactions jan+feb 2004 www. hf. faa. gov/Webtraining/ Non. Vis. Displays/n 08 c_Olfactory. htm European Telecommunications Standards Institute's Human Factors; Guidelines the multimodality of icons, symbols and pictograms v 1. 1. 1: Olfactory Guidelines

Symbolic Olfactory Display: olfactory icon Olfactory Icon: scent output to display information where scent is environmentally and semantically linked to the information conveyed. (ie: gunpowder-scented Quake)

Symbolic Olfactory Display: smicon: scent used to convey information that has only an abstract relationship with the information it expresses. (ie: incense clock) smicon Incense pattern of the “Hundred Gradations Incense Seal” of Shen-Li (1073 AD), from Bendini, S. A. (1964) Holy Smoke: The Oriental Fireclocks. New Scientist 21: 537 -539.

Section 5: A Future for Computer Controlled Scent Ambient Media/ Calm Computing Commercial, Gaming, Learning Ubicomp output

Thank you. Joseph 'Jofish' Kaye jofish@cornell. edu www. jofish. com/talks/

By request: highly recommended books mentioned in the talk. Emperor of Scent by Chandler Burr - a biography of Luca Turin Perfume by Patrick Suskind - a novel Aroma: The Cultural History of Smell by Constance Classen et. al. - a history of smell in different cultures Computation & Human Experience by Philip Agre why and how to question your field's core assumptions and metaphors. Many more references at jofish. com/thesis/reflist. html