Academic Apps

The academic app store

Ian Thornton

Psychology Department, Swansea University

Thursday, 13 September 12

ECVP 2012 Academic Apps Ian Thornton

Take home message

• Mobile apps have revolutionised the software industry

• “Academic apps” could dramatically change the way we produce and consume science



Take home message

• Produce -- mobile devices can be useful experimental platforms

• Produce -- academic apps could allow us to “market” our science in new ways

• Consume -- readers vs. users



iPad Experiments

• The iPad (and similar mobile devices) have potential as powerful experimental tools

• Easy to set-up and run a range of experiments (particularly with platforms such as Unity)

• Flexible ways to measure performance



• Changes the dynamics of traditional, lab-based, screen-and-keyboard tasks

• Potential for mass, remote data collection

• Easy to network, for joint action

iPad Experiments



Take home message

• Produce -- mobile devices can be useful experimental platforms

• Produce -- academic apps could allow us to “market” our science in new ways

• Consume -- readers vs. users



Pages

Keynote

MILO

GaborPlay

GoodReader

Visual Acuity XL



Overview

• Background on “apps”

• Academic Apps -- a working example

• Discussion



Background

• Mobile apps are single-use or limited function pieces of application software, designed for smart phones and tablets.

• They redefined “software” as “content”

• The majority of apps are still games, but also span the whole range covered by traditional applications (500, 000 titles just on Apple Store)



Background

• Reduction in scale/function initially driven by tech limits of mobile devices. Now desktop apps

• App Store (2008) opened up a whole new way to market and distribute software

• Independent developers vs. software giants



Overview

• Background on “apps”

• Academic Apps -- a working example

• Discussion



Academic App

Deliver scientific content by merging the concepts of an advanced e-book reader with a (fully) functional experimental task

Copyright�2004�Psychonomic�Society,�Inc. 38

Perception�&�Psychophysics2004,�66�(1),�38-50

This�article�introduces�a�new�task�for�exploring�the�se-quential�selection�of�multiple�target�items�during�search-like�behavior.�This�multi-item�localization�(MILO)�task�dif-fers�in�a�number�of�respects�from�traditional�visual�searchparadigms�and,�in�particular,�places�a�strong�emphasis�onthe�temporal,�as�well�as�the�spatial,�aspects�of�behavior.�Wewill�begin�by�describing�the�novel�features�of�the�MILOtask�and�will�discuss�some�of�the�theoretical�considerationsthat�gave�rise�to�them.�Next,�we�will�present�three�experi-ments�that�demonstrate�how�it�can�be�used�for�measuringhuman�performance.�We�will�conclude�by�discussing�theease�with�which�the�MILO�task�can�be�adapted�in�order�toaddress�a�wide�range�of�empirical�questions.�

The�main�features�of�the�MILO�task�are�illustrated�inFigure 1.�Observers�are�first�presented�with�a�trial-specific

alphabetical�sequence�of�target�items�(e.g.,�the�letters�E,�F,G,�and�H).�This�sequence�is�then�distributed�within�a�fieldof�four�additional�distracting�items,�two�preceding�and�twofollowing�the�target�sequence�(e.g.,�the�letters�C,�D,�I,�and�J).The�observers’�task�is�to�locate�each�target�in�alphabeticalorder�and�to�select�it,�using�the�mouse.�Each�trial�thus�yieldsN reaction�times�(RTs),�where N�refers�to�the�length�of�thetarget�sequence.�The�experiments�reported�below�all�makeuse�of�such�four-item�alphabetic�sequences.�However,�it�isimportant�to�point�out�that�the�task�parameters—that�is,sequence�type�(i.e.,�alphabetic,�numeric,�or�arbitrary),�se-quence�length,�sequence�complexity�(i.e.,�interitem�rela-tionships),�the�nature�and/or�number�of�distracting�items,the�ease�of�physically�selecting�items�(e.g.,�size�or�proxim-ity�of�items),�and�so�on—can�easily�and�systematically�bevaried. A�constant�stimulus�set�has�been�used�throughoutthis�article,�since�our�initial�goal�is�to�introduce�the�basicstructure�and�more�general�applications�of�the�task.

The�MILO�task�clearly�has�its�roots�in�the�well-known�vi-sual�search�paradigm.1�In�visual�search,�observers�are�askedto�report�on�some�property�of�a�designated�target�item�(e.g.,location�or�presence/absence)�presented�among�a�variableset�of�distracting�items.�Such�tasks�are�typically�designedso�that�each�trial�isolates�a�single�act�of�selection�in�theface�of�competition�from�the�distracting�items,�measuringa�single�response.�In�this�way,�we�have�learned�a�great�dealabout�the�basic�building�blocks�of�vision,�the�overall�orga-nization�of�visual�information�processing,�and�in�particu-lar,�the�relationship�between�vision�and�attention�(e.g.,�Dun-

The�authors�contributed�equally�to�this�research,�and�authorship�orderwas�determined�arbitrarily.�Thanks�are�extended�to�Richard�Abrams,�Ray-mond�Klein,�Cathleen�Moore,�Jay�Pratt,�Kimron�Shapiro,�Jeremy�Wolfe,and�an�anonymous�reviewer�for�useful�comments�and�suggestions�and�toGeorge�Alvarez,�Helga�Arsenio,�Serena�Butcher,�Megan�Hyle,�andStephan�Simon�for�assistance�with�data�collection.�Correspondence�maybe�addressed�to�either�author:�T. S.�Horowitz,�Center�for�Ophthalmic�Re-search,�Brigham�and�Women’s�Hospital,�64�Sidney�St.,�Suite�170,�Cam-bridge,�MA�02139�(e-mail:�[email protected])�or�I. M.�Thorn-ton,�Max�Planck�Institute�for�Biological�Cybernetics,�Spemannstrasse�38,72076�Tübingen,�Germany�(e-mail:�[email protected]).

Note—This�article�was�accepted�by�the�previous�editorial�team,headed�by�Neil�Macmillan.

The�multi-item�localization�(MILO)�task:Measuring�the�spatiotemporal

context�of�vision�for�action

IAN�M.�THORNTONMax�Planck�Institute�for�Biological�Cybernetics,�Tübingen,�Germany

and

TODD�S.�HOROWITZBrigham�and�Women’s�Hospital,�Boston,�Massachusetts

and�Harvard�Medical�School,�Boston,�Massachusetts

We�describe�a�new�multi-item�localization�task�that�can�be�used�to�probe�the�temporal and�spatialcontexts�of�search-like�behaviors.�A�sequence�of�four�target�letters�(e.g.,�E,�F,�G,�and�H)�was�presentedamong�four�distractor�letters.�Observers�located�the�targets�in�order.�Both�retrospective�and�prospec-tive�components�of�performance�were�examined.�The�retrospective�component�was�assessed�by�hav-ing�target�items�either�vanish�or�remain�once�they�had�been�located.�This�manipulation�had�little�effecton�search�performance,�suggesting�that�old�target�items�can�be�efficiently�ignored.�The�prospectivecomponent�was�assessed�by�shuffling�future�target�and�distractor�locations�after�each�response.�This�ma-nipulation�revealed�that�observers�typically�plan�ahead�at�least�one�target�into�the�future.�However,�evenwhen�observers�cannot�plan�ahead,�they�are�still�able�to�ignore�old�targets.�These�findings�suggest�thatboth�“what�you�did”�and�“what�you�intend�to�do”�can�influence�the�localization�and�selection�of�targets.

0!

0.2!

0.4!

0.6!

0.8!

1!

1.2!

1.4!

1! 2! 3! 4! 5! 6! 7! 8!

Seri

al R

eact

ion

Tim

e (s

ecs)!

Target Number!

Shuffle! Baseline!



Discussion/FAQs• Why not just use “web apps”?

• Is it expensive/complicated to write apps?

• Giving away the goods? (Freemium model?)

• Ad Hoc or Academic App store?

• Who pays?

• Is this really “Open Science?”


Technology

Academic Apps