Visual control of an action discrimination in pigeons · Visual control of an action discrimination in pigeons Muhammad A. J. Qadri $ Department of Psychology, Tufts University, Medford,

Visual control of an action discrimination in pigeons

Muhammad A J Qadri $Department of Psychology Tufts University

Medford MA USA

Yael AsenDepartment of Psychology Tufts University

Medford MA USA

Robert G CookDepartment of Psychology Tufts University

Medford MA USA

Recognizing and categorizing behavior is essential for allanimals The visual and cognitive mechanisms underlyingsuch action discriminations are not well understoodespecially in nonhuman animals To identify the visualbases of action discriminations four pigeons were testedin a gono-go procedure to examine the contribution ofdifferent visual features in a discrimination of walkingand running actions by different digital animal modelsTwo different tests with point-light displays derived fromstudies of human biological motion failed to supporttransfer of the learned action discrimination from fullyfigured models Tests with silhouettes contours and theselective deletion or occlusion of different parts of themodels indicated that information about the globalmotions of the entire model was critical to thediscrimination This outcome along with earlier resultssuggests that the pigeonsrsquo discrimination of theselocomotive actions involved a generalized categorizationof the sequence of configural poses Because the motorsystems for locomotion and flying in pigeons share littlein common with quadruped motions the pigeonsrsquodiscrimination of these behaviors creates problems formotor theories of action recognition based on mirrorneurons or related notions of embodied cognition Itsuggests instead that more general motion and shapemechanisms are sufficient for making suchdiscriminations at least in birds

Introduction

The detection recognition categorization andinterpretation of the behavior of other animals arevital to the survival of many species An essentialsocial skill in humans our capabilities for interpretingbehaviors are highly developed and potentially spe-cialized In the last decade a marked upsurge in

research examining action recognition in humans hasbeen inspired in part by the discovery of mirrorneurons in monkeys (Buccino Binkofski amp Riggio2004 Decety amp Grezes 1999 di Pellegrino FadigaFogassi Gallese amp Rizzolatti 1992) This has givenrise to the development of a number of motor-basedtheories of human action embodied cognition lan-guage intentionality and social cognition (Arbib2005 Engel Maye Kurthen amp Konig 2013 Gallese2007 Grafton 2009 Iacoboni 2009 Jeannerod 2001Rizzolatti Fogassi amp Gallese 2001 Wilson ampKnoblich 2005) One recent idea for instance hascentered on the notion that humans have an actionobservation network that is critically tied to theembodied simulation of the movements of others andis essential to understanding conspecific actions andintentions (eg Grafton 2009) It has been suggestedthat this system uses species-specific motor-basedknowledge to recognize actions by internally simulat-ing or emulating them (Buccino Lui et al 2004Jeannerod 2001 Wilson amp Knoblich 2005)

Recognizing and interpreting the behavior of bothconspecifics and heterospecifics is equally critical fornonhuman animals serving important functions incourtship mate selection communication territorydefense learning by imitation and social foraging(Byrne amp Russon 1998 Fernandez-Juricic Erichsen ampKacelnik 2004) Not all animals have the luxury of theanalytical power in a human brain however Graftonrsquos(2009) proposed human action observation networkfor example exceeds the total brain size of many birdswhose neural hardware has likely been limited in size bythe evolutionary demands of muscle-powered flightTherefore understanding how action recognition andits neural mechanisms work in other nonhuman speciesmay provide important insight into our own actionrecognition The theoretical analysis of action recog-

Citation Qadri M A J Asen Y amp Cook R G (2014) Visual control of an action discrimination in pigeons Journal of Vision14(5)16 1ndash19 httpwwwjournalofvisionorgcontent14516 doi10116714516

Journal of Vision (2014) 14(5)16 1ndash19 1httpwwwjournalofvisionorgcontent14516

doi 10 1167 14 5 16 ISSN 1534-7362 2014 ARVOReceived August 5 2013 published May 30 2014

nition by nonhuman animals has progressed muchmore slowly because of the difficulty of controlling andusing lsquolsquobehaviorrsquorsquo in experimentally analytic situationsAnimals just donrsquot take direction well Digital softwareused to create animated displays of behavior howeverholds considerable promise for moving beyond thisproblem

Recently we successfully taught pigeons to discrim-inate and group the walking and running actions ofeight different digital animals using life-like articulat-ed animated models in a gono-go task (Asen amp Cook2012) Because these locomotor activities are likelysalient natural action categories (Malt et al 2008) theyprovided a good starting place for building on the priorresults of video-based action recognition (Dittrich ampLea 1993 Dittrich Lea Barrett amp Gurr 1998Jitsumori Natori amp Okuyama 1999) In that studyeach digital animal model ran or walked in place on atextured background (see examples in Figure 1) Toencourage action categorization the digital modelswere rendered from 12 different camera perspectives(combinations of elevation azimuth and distance) Itwas found that (a) this type of action discriminationwas easily acquired (b) it showed significant transfer tonovel species moving in biologically appropriate butdistinct ways (c) it exhibited viewpoint invariance overcamera distance elevation and perspective (d) it didnot vary substantially with variations in presentationspeed (e) and it showed selective interference with theinversion of the video or the randomization of itssequential frames The results seemed most consistentwith the hypothesis that the pigeons learned actioncategories for the different behaviors as a series of

sequenced poses Given that pigeon locomotion likelydoes not share motor representations in common withthe different quadruped actions that they were dis-criminating these results suggest that behaviors cansometimes be visually discriminated without theirembodiment in the observer

Computational models have explored the problemof human behavior recognition based exclusively onvisual information for a variety of functions (Aggar-wal amp Cai 1999 Poppe 2010 Wang Hu amp Tan2003) One computer vision approach focuses on thehigher-level global or configural organization amongdifferent body parts to recognize action The repre-sentation used in these theories often involveshierarchical geometry-based configural models cod-ing the relative motion of body limbs and joints(Aggarwal amp Cai 1999) A second approach codesnonconfigural and often nonparametric representa-tions to sufficiently discriminate among behaviorsThese theories vary in many ways including how andwhat information is encoded from global represen-tations such as space-time volumes or integratedsilhouettes to more localized features such as opticflow or periodic motion trajectories (eg Bobick ampDavis 2001 Polana amp Nelson 1997 Schindler amp VanGool 2008) One nonconfigural account of the actionsin Figure 1 for example might isolate the localizedmovement of the five different points traced in eachexample Given any one of these paths but especiallythose of the feet it would be possible to determine ifthe model were running or walking without processingthe entire figure Some models of human biologicalmotion perception utilize both configuraltop-down

Figure 1 Example of one of the eight animal models used in these experiments to exemplify the actions of walking and running It is

shown as rendered from a low close side perspective Superimposed on the displays are the different motion paths of five body

parts (nose neck junction tail junction fore right foot and rear right foot) These paths were not present in the stimuli tested with

the pigeons

Journal of Vision (2014) 14(5)16 1ndash19 Qadri Asen amp Cook 2

and localbottom-up information with some success inreproducing experimental outcomes (Giese amp Poggio2003) It is not possible to precisely discriminateamong the wide variety of proposed computer modelsbut a key feature in these methods is the use of globalor local cues Identifying which cues pigeons usewould help to determine the visual and cognitivemechanisms involved in avian action recognition andin identifying the relevant class of computationalmodels for comparison

Asen and Cookrsquos (2012) evidence seemed to favorthe pigeonsrsquo use of the more generalized higher-levelorganization of the digital modelsrsquo actions Firstinverting the video disrupted the birdsrsquo discriminationmuch as in humans (Dittrich 1993) Such stimulusinversions only minimally disturb the localized non-configural features of the displaysmdashalthough effectslike this have been attributed to spatially relevant localmotion detection (Hirai Chang Saunders amp Troje2011) Second the gaits of the animated animalmodels varied greatly (ie ponderous elephants lithecats) but each of the models supported good transferof the locomotion discrimination suggesting a higher-level recognition of the actions than species-specificfeatures Finally the viewpoint invariance of thepigeonsrsquo discrimination across perspectives suggeststhat the precise appearance of the features or theirrelationships was not particularly critical Althoughthese results suggest that the pigeons use larger-scaleor global features the question can be investigateddirectly and more precisely using digitally alteredstimuli

The present experiments specifically investigated therepresentation of these locomotive actions in thepigeons by testing them with different display manip-ulations The goal of the experiments was to determineif the pigeons relied more on global or local informa-tion to discriminate the actions of the digital modelsPigeons previously trained to discriminate walking andrunning actions were tested In Experiment 1 weinvestigated whether the pigeons could transfer theirestablished action discrimination to point-light displays(PLDs) similar to those studied in human tests oflsquolsquobiological motionrsquorsquo (Johansson 1973) In Experiment2 the pigeons were tested with models in which internallocal features were eliminated by using only the contouror silhouette of the acting models Finally in Exper-iment 3 different portions of the models were digitallyoccluded or deleted to determine which parts of themodels were critical in mediating the discriminationBetter understanding how the pigeons processed andclassified precisely controlled movements that simulatethe actions of different animals may offer insights intohow they represent behaviors generally and the natureof the visual mechanisms involved

Experiment 1

In humans lsquolsquobiological motionrsquorsquo displays are regu-larly used to examine the discrimination and recogni-tion of actions (Blake amp Shiffrar 2007 Johansson1973) Consisting of coordinated moving points or dotscorresponding to the articulated motions of differentbehaviors such PLDs powerfully invoke the perceptionof a behaving actor in humans Humans can easilyclassify a wide variety of actions and recognize manysocially relevant features (eg age gender emotion)from these simple moving elements (Blake amp Shiffrar2007) Thus PLD perception seems to require thespatial and temporal integration of the separated anddiscrete elements into a global perception of actionwithout corresponding form information (however seeThirkettle Benton amp Scott-Samuel 2009) Important-ly this perception is derived from both global and localfeatures of the stimuli (Beintema amp Lappe 2002 Hiraiet al 2011)

This ease of recognizing biological motion in PLDstimuli by humans has in turn generated sharp interestin whether such displays similarly generate the sametype of perception in nonhuman mammals (Blake1993 J Brown Kaplan Rogers amp Vallortigara 2010Oram amp Perrett 1994 Parron Deruelle amp Fagot2007 Puce amp Perrett 2003 Tomonaga 2001) and birds(Dittrich et al 1998 Regolin Tommasi amp Vallorti-gara 2000 Troje amp Aust 2013 Vallortigara Regolinamp Marconato 2005) For birds the results have beenmixed

Dittrich et al (1998) trained pigeons to discriminatebetween pecking and nonpecking behaviors of conspe-cifics using video playback Following fully detailedvideo training those birds that learned the discrimi-nation showed limited transfer to PLDs of the samebehaviors A follow-up experiment revealed that fourof eight naıve pigeons could learn to discriminate PLDdisplays of these same behaviors but they showed notransfer of the discrimination to fully detailed videosThese mixed results suggest that the processing of thePLDs by pigeons does not readily generate the samepercept of a behaving animal as expressed via videoplayback

Troje and Aust (2013) recently trained eight pigeonswith PLDs in a direction discrimination in which theyhad to discriminate left-walking from right-walkinghuman or pigeon PLD walkers in a choice task Theauthors then tested the discrimination with globallyand locally inconsistent displays and various inversioncontrols They found two pigeons that appeared to beresponding to the globally facing direction of thewalkers and the remaining six attended primarily tothe dots corresponding to the movement of the feetThe latter pattern of results suggests that for themajority of the pigeons their perception and discrim-

ination of these biological motion displays was locallybiased

Young chicks have been tested several times withbiological motion animations (Regolin et al 2000Vallortigara et al 2005) Using an imprinting proce-dure Regolin et al (2000) imprinted a large number ofchicks on PLD stimuli of either a walking or ascrambled hen When later tested for preference withboth displays females displayed a slightly greaterpreference for the imprinted animations (walking orscrambled) and males showed a small amount ofavoidance to the imprinted displays (walking orscrambled) Using a preferential proximity paradigmVallortigara et al (2005) investigated PLD perceptionin newly hatched chicks by examining their distancefrom two possible test displays Testing a large numberof chicks their experiment revealed a small butsignificant proximity preference for an articulated henPLD an articulated cat PLD and a scrambled henPLD when compared to random or rigid dot motiondisplays While it is not clear that these displays areperceived exactly as intended the authors suggest thesefindings imply that chicks have an innate predispositionfor processing the types of features that underliebiological motion perception

In the current experiment we created PLD stimulithat retained the articulated structure and motionfeatures of our already established walking and runningdigital models We then examined how discriminationwith these articulated PLD stimuli compared to thediscrimination of full-figured stimuli and the discrim-ination of several important controls One of thesecontrols was a scrambled condition which hadidentically moving dots positioned randomly about thedisplay This control contains all of the same motioninformation but lacks the structural articulationcoordination and coherence of the motion thatpromotes biological motion perception The secondwas an inversion condition in which the dot pattern wasinverted resulting in the lsquolsquolegsrsquorsquo of the PLD modelpointing up and the lsquolsquoheadrsquorsquo and lsquolsquotorsorsquorsquo positionedtoward the bottom This control provides the samecoordinated articulation and periodic timing butdisrupts location-specific motion features (eg as inBlake amp Shiffrar 2007 Troje amp Aust 2013) Typicallythis control has been interpreted as disrupting globalprocessing but recent research shows that this arguablyglobal manipulation affects the weight given to localmotion cues during PLD displays (Hirai et al 2011)Thus if inversion disrupts otherwise capable PLDdiscrimination it would need to be determined if thedisruption resulted from an interaction of local andglobal properties Last we tested a randomized framecondition in which all of the same frames werepresented as in the normally articulated PLD stimulusbut in a random sequence This condition disrupts

motion-based cuing while retaining the same staticframes during presentation (Asen amp Cook 2012 Cookamp Roberts 2007 Koban amp Cook 2009) Anydiscrimination of this condition suggests that thecoherent pattern of motion and the form features areirrelevant and that some static cue such as the presenceof the figure in a certain region of the display can besufficient for discrimination (Cook amp Roberts 2007)Experiment 1 consisted of two different tests of thesePLD stimuli with the pigeons Any greater degree oftransfer from the ongoing fully figured locomotiondiscrimination to the normally articulated PLD stimulirelative to the different controls would be consistentwith the hypothesis that the pigeons see the biologicalmotion in these displays like humans do

Methods

Animals

Four male pigeons (Columba livia) were tested G1G2 S3 and Y4 They were maintained at 80ndash85of their free-feeding weights with free access to grit andwater These pigeons were already trained to discrim-inate walking and running actions and did not requireany additional training All procedures were approvedby the Tufts University Internal Animal Care and UseCommittee which adheres to ARVO guidelines

Apparatus

Testing was conducted in a computer-controlledchamber Stimuli were presented on an LCD monitor(NEC Accusync 51VM 1024 middot 768 60-Hz refresh rate)recessed 8 cm behind a 33 middot 22 cm infrared touchscreen(EZscreen EZ-150-Wave-USB) A 28-V ceiling lightwas illuminated at all times except during time-outs Acentral food hopper (Coulbourn Instruments) underthe touchscreen delivered mixed grain

Procedure

Gono-go discrimination testing Each trial was initiatedby a peck to a centrally presented 25-cm white readysignal This signal was replaced by a video of a digitalmodel animal started from a randomly selected frameand repeatedly looped from there for 20 s Two pigeonswere reinforced for pecking at lsquolsquorunningrsquorsquo models andtwo for pecking at lsquolsquowalkingrsquorsquo models (described below)Pecks during these correct Sthorn actions were reinforcedwith 29-s access to mixed grain on a variable intervalschedule (VI-10) so that a single peck would result inreinforcement with uniform probability from 0 to 20 safter the peck occurred An additional 29-s reward wasprovided at the end of Sthorn presentations Pecks to theincorrect S action resulted in no reward and a variable

dark time-out at the end of the presentation (05 s perpeck for G1 1 s per peck was used) During baselinetrials a small percentage of Sthorn trials were randomlyselected to be probe trials during which no reinforce-ment was delivered These trials allowed for theuncontaminated measurement of the positive peck ratewithout the interruption or signaling of the foodpresentations All baseline Sthorn-dependent measureswere calculated from these probe trialsBaseline The digital stimuli were 115 middot 115 cmcompressed AVI (using Microsoft Video1 compression)videos of three-dimensionally rendered animal modelsthat were running or walking in a continuous loop Thestimuli were created and rendered with 3-D figuralanimation software (Poser 7 and 8 Smithmicrocom)using third-party models of the animals and theiractions (Daz 3D wwwdaz3dcom and Eclipse Studioswwwes3dcomindex2html) Six animal models avail-able from prior training were used in the baseline buckcamel cat dog elephant and human (seeSupplementary Movies 1ndash6 for examples with the dogand buck) The dog and buck action models were alsorendered with two different skins bringing the totalnumber of training lsquolsquoanimalsrsquorsquo to eight

Using different biomechanical motion models char-acteristic of the species depicted each animal modelmoved in a fixed central position (ie walking orrunning lsquolsquoin placersquorsquo) The number of frames and theirpresentation rate (frames per second) varied accordingto the digital model and action Across the models usedin the baseline set the lsquolsquorunningrsquorsquo stimuli appropriatelycycled faster (Mfrac1417 behavioral cycles per second cps)than the lsquolsquowalkingrsquorsquo stimuli (M frac14 82 cps)

All model animals were rendered from a combina-tion of six camera directions (body focus sidefrac14 08frontfrac14458 rearfrac14thorn458 and direction left-facing andright-facing) two camera elevations (low 558 andhigh 2638 relative to the surface) and two cameradistances (close far) Visual angle was calculated usinga viewing distance of 85 cm to accommodate the 8 cmthat the screen was recessed and 05 cm for the pigeonrsquosviewing distance of the stimuli at its closest point Thisyields a horizontal visual angle of approximately 268 to408 depending on the model in the close perspectiveand 88 to 128 in the far perspective

Each digital animal was illuminated from a fixedoverhead light source and rendered in one of twocontexts One context was the receding green-texturedflat lsquolsquogroundrsquorsquo surface below a pale blue lsquolsquoskyrsquorsquo used inAsen and Cook (2012) and all possible combinationsof animals and perspectives were shown in this contextThe second context contained no ground so the modelwas surrounded by just pale blue sky bounding box(RGB valuefrac14 [191 252 252]) This no-ground contexthad only been added to training in order to familiarizethe pigeons with the context used to test the transfer

stimuli so only a restricted set of stimulus configura-tions were displayed in this context the dog and buckmodels in the low close side perspective

A total of 192 different videos of each of the twoactions on the grass context were thus used in baseline(eight digital animals middot 24 perspectives) Four videos ofthe buck and dog in the no-ground context wereincluded to acclimate the pigeons to the no-groundcontext used in the tests all from the low close sideperspective Prior to the experiment the four pigeonswere very good at discriminating these actions acrossall of these models perspectives and contexts

Baseline sessions consisted of 84 trials (42 walking42running) The animal camera distance and elevationvaried randomly for 72 trials but equivalent counts ofcamera perspective (canonical side three quartersfront three quarters rear but not facing direction) werepresented Of the Sthorn trials in this set 15 weredesignated as nonreinforced Sthorn probes as describedabove Further 12 additional no-ground trials (sixwalkingsix running) were randomly mixed into this setdepicting the dog or buck from the right-facing low andclose perspectivesPLD test 1 The PLD stimuli were created by placing12-mm (88) flat black dots at the key joints of themodels in the digital software These dots moved in acoordinated fashion in the same positions as themodelsrsquo joints PLD stimuli were created for both thebuck (27 dots five per limb four for the torso two forthe neck and one for the head see SupplementaryMovie 7) and dog (28 dots four per limb five for thetorso four for the tail two for the neck and one for thehead) models To control for interstimulus variabilityonly these two stimuli were manipulated throughoutthese experiments To maximize the visibility of thedots the ground context was omitted and the stimuliwere rendered from the low close side cameraposition The resulting articulated PLD lsquolsquofiguresrsquorsquosubtended a visual angle of 258 to 408 matching thefully detailed animations for overall spatial extent Onecomplete cycle of the PLD displays contained the samenumber of frames as the fully figured displays (buckrunning 19 frames walking 33 frames dog running16 frames walking 50 frames) and the stimuli werepresented at the rate of 30 ms per frame (333 framesper second) again matching baseline values

For comparison three control conditions weretested The inverted control consisted of a 1808 rotationof the articulated PLD stimuli The scrambled controlhad the dots randomly positioned in the videoseliminating their configural motion but otherwisehaving each individual dot moving along the same localpathway and temporal synchrony as in the articulatedcondition As this condition was generated off-lineonly one scrambled version of each model was testedLast a randomized frame condition was tested Here

the frames of the articulated PLD stimuli wererandomly scrambled during their presentation break-ing up its coherent motion The randomized order ofthe frames was changed for each presentation but fixedfor the duration of the presentation

The four PLD conditions were tested as both walkand run actions in each session A testing sessionconsisted of the same 72 ground-context trials as inbaseline sessions two no-ground context trials with thetested animal from the low close side perspective andeight transfer trials testing the above experimentalconditions yielding 82 total trials The eight transfertrials and the two no-ground trials were all tested asnonreinforced probes A total of 10 sessions testing thebuck and dog PLD models five times each wereconductedPLD test 2 Following the unsuccessful transfer of thediscrimination in the previous test (see Results) wecreated and tested new PLD stimuli that might bettersupport the perceptual grouping of the dots For thisgoal two properties were manipulated First dots thatwere two (24 mm 168) and three times (36 mm 248see Supplementary Movie 8) larger than in the priortest were added to reduce the distances between displayelements Second the overall size and visual angle wasreduced by shifting the perspective to be roughly twotimes farther from the model (see SupplementaryMovies 9 and 10 overall visual angle 138 to 258 maxdot size 148) To reduce the number of conditions onlythe buck model was used in this test and only theinverted and randomized frame control conditions wereincluded Each test session present the six PLD transferstimuli (walking and running articulated inverted andrandomized frames) for a fixed distance (close and far)and dot size (12 mm 24 mm and 36 mm) The sixtrials testing these transfer conditions were conductedas nonreinforced probes and they were randomlymixed into a session with 72 ground-context baselinetrials and two no-ground buck trials yielding 80 totaltrials per test session Six sessions were required to testall combinations once comprising a single experimen-tal block Three blocks of testing were conductedtotaling 18 test sessions The order of tests within ablock was randomizedMetrics The LED touchscreen in the chamber used forthese experiments was unusually sensitive picking upsmall differences among pecks but also chest andfeather entries Although we will continue to refer tomeasuring peck responses the peck counts morehonestly reflect the degree of total activity directed bythe pigeons toward the displays The primary depen-dent variable analyzed was discrimination ratio (DR)the proportion of total pecks that occurred during anSthorn stimulus (ie Sthornpecks [S pecksthornSthornpecks]) Thisadjusts for each birdrsquos individual rate of respondingand scales nicely from 0 to 1 such that 5 is chance

performance and 1 is perfect discrimination To bestillustrate the birdsrsquo discrimination as it relates to theDR the reported peck rates have been adjusted foreach birdrsquos base rate of pecking by normalizing eachbirdrsquos data to the total average pecking to all positivebaseline trials The baseline peck rates reported andanalyzed however concern only the suitable compar-ison stimuli (ie the same perspective as test stimuli)Consequently their values are typically near 1 as aresult of the normalization but they are not fixed there

Results

Baseline All four pigeons were very good at discrim-inating the different actions of all eight animal modelson the baseline trials of the first experiment exhibitinggreater pecking to the Sthornaction (mean pecksfrac14810 SEfrac14 161) than the S action (mean pecks frac14 154 SE frac1423) Mean discrimination ratio (see Metrics) whencomputed for each bird individually and then averagedtogether was 083 (SEfrac14 005) A one-sample t testconfirms that this is significantly above the chance levelof 05 t(3)frac14 69 pfrac14 0006 d frac14 34 (all p values and ttests reflect two-tail comparisons alphafrac14 005 for thisand all comparisons) This action discrimination wassignificantly affected by the model animal displayingthe action as analyzed using a one-way repeated-measures ANOVA on DR F(5 15)frac14 108 p 0001g2p frac14 0783 which paired comparisons indicate iscaused by a lower DR with the elephant model Onesample t tests comparing DR to chance for each modelseparately confirmed however that all models sup-ported significantly above-chance levels of discrimina-tion of the two actions ts(3) 37 ps 0032 ds 18 This action discrimination was also invariantacross azimuth elevation and distance Becauserandomized and combinatorial complexity in thebaseline displays resulted in untested combinations offactors separate one-way repeated-measures ANOVAs(azimuth without facing direction elevation distance)were used to separately analyze these different aspectsof perspective No effects of these factors were foundThese results match those previously reported for thesesame birds (Asen amp Cook 2012) in which cameraperspective also had no influence on discriminationPLD test 1 The pigeons showed no capacity todiscriminate among the actions when depicted as PLDstimuli despite their continued and excellent discrimi-nation of complete models Shown in Figure 2 are thenormalized peck rates for Sthorn and S actions for thematched baseline and different PLD test conditionsBaseline DR was computed from all comparable trialstesting the buck and dog models from the perspectiveconfiguration used for the PLD stimuli and it wassignificantly above chance t(3)frac14 102 pfrac14 0002 d frac14

51 Importantly there was no significant discrimina-tion of the articulated PLD stimuli where the dotsmirrored the motions of complete figures t(3)frac14 002Given the lack of discrimination in the articulatedcondition it is perhaps not surprising that none of thePLD controls supported significant discriminationeither ts(3) 15

We also examined the results for each pigeonseparately to determine if any individual bird may haveperceived the actions within the PLD stimuli Similaranalyses were conducted using session as the repeatedfactor for each bird but the results were identical to thegroup analysis Each bird individually discriminatedthe complete models ts(9) 85 ps 0001 but notthe articulated PLD condition or different controlsts(9) 22PLD test 2 This second test also failed to reveal anyevidence of discrimination mediated by the PLDstimuli despite modifications to better support per-ceptual grouping and configural perception in thedisplays Table 1 lists the mean normalized pecking forthe baseline and PLD test conditions The baselinestimuli continued to support excellent discriminationwith both close and far versions of the completemodels ts(3) 90 ps 003 A repeated-measuresANOVA (distance middot dot size) of pecks to thearticulated PLD stimuli revealed no significant effectsSimilar analyses of performance with the PLD controlsalso found no evidence that the various conditionsaffected discrimination Further one-sample t testssuggest no discrimination was found for any control ordisplay condition ts(3) 21 except for the randomframe condition t(3)frac1433 pfrac14 0044 Post-hoccorrections for these multiple comparisons using the

Holm-Bonferroni method indicated that this was likelynot a significant result Again analyses of theindividual birds with block as the repeated factor alsofailed to find evidence of discrimination in theseconditions

Discussion

Pigeons trained to discriminate the walking andrunning actions of a wide variety of complete fullyfigured articulated models showed no capacity totransfer this discrimination to PLD stimuli corre-sponding to these actions and models This was foundin two different tests during Experiment 1 Althougheasily discriminated by humans this type of lsquolsquobiologicalmotionrsquorsquo display failed to support the discrimination of

Figure 2 Mean normalized peck rates for the four pigeons in Experiment 1 tested with different types of PLDs Error bars indicate the

standard error of each condition

Condition Dot size

Close Far

Sthorn S Sthorn S

Baseline 108 013 090 018

Articulated 1middot 009 009 007 010

2middot 004 004 009 006

3middot 013 010 012 005

Inverted 1middot 009 010 010 009

2middot 008 005 009 008

3middot 014 008 009 010

Randomized 1middot 006 012 009 009

2middot 006 010 011 006

3middot 010 013 008 010

Table 1 Mean normalized pecks in PLD condition during secondtest of Experiment 1

these well-trained actions in pigeons This was trueacross considerable variations in the size of the definingdots and the visual angle of the display that attemptedto promote the perceptual grouping of the separatepoints

There are several possible reasons for this difficultyFirst the overall appearance of the PLD stimuli andthe complete models are different One of the reasonsfor testing PLDs is that they make it possible toexamine the independent contribution of articulatedand coordinated motion by eliminating form informa-tion The resulting large alteration in the forminformation (model to dots) however may have causeda degree of neophobic nonresponding Supporting thishypothesis three of four pigeons in this experimentpecked less to the PLD stimuli than full-featureddisplays and furthermore Dittrich and colleaguesrsquo(1998) pigeons similarly showed reduced levels ofpecking during their PLD transfer Such loweredreduced pecking suggests some degree of generalizationdecrement related to unfamiliarity likely contributes tothe poor performance of the pigeons with biologicalmotion stimuli

Another potentially important reason for the pi-geonsrsquo inability to discriminate PLDs is that thesedisplays require the perceptual grouping of widelyseparated and disconnected points into a unifiedconfiguration In this case the lack of connected form-based cues may prevent the activation of the motioncues required for the discrimination Pigeons havefrequently exhibited problems grouping separatedelements into larger wholes (Lea Goto Osthaus ampRyan 2006) With hierarchically arranged stimuli theyfrequently show a bias to initially process localelements over global ones (Cavoto amp Cook 2001) Intheir study of PLD direction perception Troje andAust (2013) found the majority of pigeons exhibited alocal bias Correspondingly pigeons also have hadtrouble detecting the larger global structure of Glassdot patterns completing separated amodal displaysand the larger symmetry of line-based patterns (Huberet al 1999 Kelly Bischof Wong-Wylie amp Spetch2001 Sekuler Lee amp Shettleworth 1996)

Although pigeons can detect and group globalpatterns under the right conditions (Cook 2001Cook Goto amp Brooks 2005) this appears to emergewith experience or secondarily to initial attention tolocal elements A similar local bias has also beensuggested about the visual cognition of humanindividuals diagnosed to be on the autism spectrumThis may possibly be the reason for their increaseddifficulty in detecting biological motion actions andsocial information in PLDs (Kaiser amp Shiffrar 2009)While some animal studies of biological motion havefound more intriguing results than these with PLDstimuli our results are also part of a general trend that

pigeons and perhaps other animals just do not findthe coordinated actions in such dotted stimuli as easyto perceive as humans (Dittrich et al 1998 Troje ampAust 2013)

Experiment 2

Experiment 2 examined if the pigeons required morecomplete or connected form information to detect themotion patterns that they use to discriminate theseactions To explore how varying types of forminformation contributed to the discrimination wetested contour-only and silhouette versions of theanimal models (examples included in Figures 3 and 4see Supplementary Movies 11 and 12) These stimuliretained the global motions of the models using anexterior contour that was completed and connectedwhile concurrently reducing interior texture and forminformation

Several previous studies have examined the use ofthis information for object perception in pigeons Theyhave suggested that pigeons are able to interpretsilhouettes of objects at least partially correctly (egCook Wright amp Drachman 2012 Peissig YoungWasserman amp Biederman 2005 Young PeissigWasserman amp Biederman 2001) Besides being able todiscriminate objects in part based on their silhouettethis discrimination may incrementally improve ifdynamically presented in a manner consistent with therigid structure of an underlying object (Cook amp Katz1999) Tests with contour-only static stimuli havegenerally suggested that this type of stimulus is moredifficult to discriminate than silhouettes (Cabe ampHealey 1979 Cook et al 2012 Peissig et al 2005)

We conducted two separate tests first testingcontour-only models and then silhouette models Thecontour test stimuli consisted of black outlines of themodels running or walking The silhouette test stimuliconsisted of the models with their interior solidly filledin with the average color of the fully rendered modelHence these two types of stimuli removed the localinternal detail of models while retaining their globalconnected form and associated motion informationThree additional conditions were tested To evaluatethe role of temporal and spatial features controls withrotated and inverted versions of the stimuli wereincluded The role of coherent motion features in thediscrimination was again evaluated using tests in whichthe frames of the video stimuli were randomized Themain question was whether such models having bothbounded and connected form information would besufficient to support the established action discrimina-tion

Methods

Animals and apparatus

The same pigeons and apparatus were used as inExperiment 1 After the contour test G1 was nolonger tested for reasons unrelated to the experiment

Procedure

Contour test This first set of test stimuli consisted ofthe contoured outline of the buck animal model Thebaseline buck model without the ground was modified

using MATLAB The close and far low side stimuliwere used and a two-pixel (48) black contour wasgenerated at the border of the figures on a frame-by-frame basis using the close and far low side videos onthe blue background A simple edge-detection algo-rithm was used Briefly if any of the eight pixelssurrounding a nonbackground pixel was the back-ground color it was considered a border pixel and ifnone were the background color they were consideredinterior pixels All border pixels were colored black ona frame-by-frame basis and the interior pixels thattouched those border pixels were also colored black

Figure 3 Mean normalized peck rates for the four pigeons in Experiment 2 tested with different contour displays Error bars indicate

the standard error of each condition

Figure 4 Mean normalized peck rates for the three pigeons in Experiment 2 tested with different silhouette displays Error bars

indicate the standard error of each condition

These frames were then combined into an AVI video(Cinepak codec see Supplementary Movie 11)

Two types of control stimuli were also created forthis set The control stimulus in the rotated conditioncontained the contour figure rotated 908 so that thebuckrsquos head pointed upward This stimulus couldsupport a locomotion discrimination that specificallyattended to the periodic nature of the display Theother control rotated the contoured figure 1808 whichretained the original stimulirsquos nonconfigural motionalong the vertical dimension in addition to features oftiming Finally one control condition of properlyoriented but randomized frames was also tested and itcould evaluate the use of coherent motion

The contour test stimuli and the three control stimuli(four conditions middot two actionsfrac14 eight contour stimuli)for a given distance were tested as probe trialsrandomly mixed into sessions of 72 ground-contextbaseline trials and two buck no-ground trials totaling82 trials in the test sessions Two sessions were requiredto test all conditions once forming a two-session blockFour blocks of testing were conducted with onebaseline session between blocks two and threeSilhouette test The silhouette set of stimuli wasgenerated in the same way as the contour stimuli exceptthat the border and all interior pixels were colored tocreate a uniform silhouette The color of the silhouettewas the mean of the red green and blue channels of theoriginal model as averaged across all frames in thevideo (see Supplementary Movie 12) The same fourconditions as for the contour test were used uprightrotated inverted and randomized frame presentationWith a fixed distance for each session these fourconditions (analogous to above total eight silhouettestimuli) were tested as probe trials randomly mixed intosessions of 72 ground-context baseline trials and twobuck no-ground trials Two sessions were required totest all conditions once (a two-session block) and twoblocks of testing were conducted separated by onebaseline session

Results

Contour test Shown in Figure 3 are the test results forthe contour conditions as a function of mean normal-ized peck rate The four pigeons continued todiscriminate the actions with the baseline buck stimuliused for the contour test t(3)frac14 83 pfrac14 0004 dfrac14 41As shown in the peck rates to Sthorn and S stimuli thecontour-only displays seemed to support above-chancediscrimination although it was clearly reduced relativeto the baseline conditions and the average DR wasnonsignificant t(3)frac14 27 p frac14 008 This reduction andnonsignificance was partly due to Y4 failing todiscriminate among these contour stimuli because thispigeon did not peck much at these stimuli The DR of

the remaining three pigeons showed that they discrim-inated the actions of the contour stimuli t(2)frac14 75 pfrac140017 dfrac14 44 We found no differences in performanceas a function of overall visual angle

These three contour-discriminating pigeons failed todiscriminate the actions of the model in any of the threecontrol conditions (as did pigeon Y4 with its low peckrates see Figure 3) Presentations of the Sthorn and Sactions produced peck rates reflecting nondiscrimina-tion when rotated 908 inverted 1808 or the frames wererandomly scrambled One-sample t tests of DRcompared to chance indicated no discrimination waspresent for the control displays (ts(2) 1)Silhouette test Shown in Figure 4 are the test results forthe silhouette stimuli as a function of mean normalizedpeck rate for the three pigeons that were testedBaseline DR with the buck model continued to beexcellent t(2)frac14 131 pfrac14 0006 d frac14 71 The modelrsquossilhouette also supported discrimination of the actionst(2)frac14 105 pfrac14 0009 dfrac14 61 As with the contour testthe pigeons were unable to discriminate the actionsduring any of the three control conditions Consistentwith this one-sample t tests confirmed that discrimi-nation ratios for the rotated inverted and randomizedcontrol conditions were not significantly above chancelevels both when considered across birds ts(2) 17or when considered for each pigeon individually ts(3) 21 using sessions as the repeated observation foreach bird

Discussion

This experiment revealed that walking and runningmodels depicted in contour and silhouette retainedsufficient information for the majority of the pigeons todiscriminate the depicted actions The silhouettessupported discrimination at a level nearly comparableto that with the fully featured models from the sameperspectives and the contour supported a significantbut slightly reduced level of performance For bothtypes stimulus rotation eliminated the discriminationsuggesting that local features such as the localizedspeed of legs or head movement were not a critical partof the discrimination as these cues were present in bothtypes of rotations The disruption with inverted stimulisuggests that simple vertical motion of the overallfigure and timing cues from the period of the stimuli arealso insufficient for the discrimination Finally framerandomization continued to disrupt the discriminationindicating that motion coherence is needed for thesefeature-reduced stimuli

These results in conjunction with the PLD resultsfrom the previous experiment suggest several importantfacts about the pigeonsrsquo action discrimination Themost important is that the acting models likely require

connected edges or boundaries This would readilyexplain the failure of PLD stimuli in Experiment 1 Itfurther appears that internally filled boundaries facil-itate discrimination as performance with silhouetteswas better than with just contours Asen and Cook(2012) found that these pigeons were also able totransfer from one action model to another acrossvisually discriminable novel types of lsquolsquoskinrsquorsquo Whiletraditionally silhouettes are black in the present casewe used the average color of the baseline model whichmay have contributed to their continued recognitionThe exact contribution of simply filling in the figureversus the nature of coloring remains to be resolvedThe slight reduction in performance relative to thecomplete model may suggest that internal colorfeatures and texture within the models are encoded andrepresented by the pigeons (Cook et al 2012) Perhapsthe pigeons use these internal features to helpdistinguish among the modelsrsquo different limbs or partssuch as the head torso and legs The relativecontribution of these different parts is the focus ofExperiment 3

Experiment 3

The goal of Experiment 3 was to determine therelative contributions of the modelsrsquo different limbs or

body parts to the discrimination Understanding whichparts of the display are critical to performance helps toevaluate the suggestion from the previous experimentthat the pigeonsrsquo action recognition operates on moreglobal or large-scale characteristics If the pigeonscould discriminate these action stimuli without legmotion then the inversion effect was likely the result ofglobal feature disruption and not the result of expectingcritical information in the lower region of the stimuli(cf Hirai et al 2011) Thus in this experimentdifferent portions of the digital animals were madeunavailable by manipulating the visibility of selectedportions of the model across different conditions Weconducted two different tests in which we either usedocclusion or deletion to examine the pigeonsrsquo relianceon the specific components of the animal models

In the occlusion test digital lsquolsquorocksrsquorsquo were introducedand added to the scenes to obscure the visibility ofdifferent amounts of the modelsrsquo legs (see examples inFigure 5) Although changes in relative speed ofmovement within a video (Asen amp Cook 2012) and themere presence of motion in the area of the legs(Experiment 1) have proven to be insufficient tomediate the discrimination leg-related speed posi-tioning and spatial extent are the most salient carriersof walking and running information (see Figure 1) Asa result we used variably sized rocks to hide the frontthe back or both pairs of legs to further examine theirpossible contribution to the discrimination

Figure 5 Mean normalized peck rates for the three pigeons in Experiment 3 tested in different types of occlusion conditions Error

bars indicate the standard error of each condition

For a controlled comparison we tested the sameconditions with the rock placed behind the modelrsquoslegs This was done for two reasons First it is anappropriate and natural experimental control for theocclusion test Second however is that prior studieswith pigeons have found this type of lsquolsquobehindrsquorsquocondition appears to be disruptive to ongoing dis-crimination performance in several settings (eg seeDiPietro Wasserman amp Young 2002 Koban amp Cook2009 Lazareva Wasserman amp Biederman 2007) Thiseffect seems to be driven by a difficulty of decomposingnovel edge relationships that are still present when thelsquolsquooccluderrsquorsquo is in a nonoccluding position (Lazareva etal 2007) We wanted to test the reliability of thislsquolsquobehind maskingrsquorsquo effect in this different context and tohelp determine whether the movement of the modelsrsquolimbs might help to overcome these apparent process-ing difficulties by supporting better segregation of themodel from the background elements

A similar strategy was employed in the deletion testHere we digitally removed specific parts of the modelswithout altering the movement of the remaining partsby using the figural software to not render thesecomponents (see examples in Figure 6) This digitalamputation yields a similar effect as the occlusion testbut with greater precision when removing targets Forexample occluding both legs clearly disrupts thevisibility of the leg motions but it simultaneouslydeprives the pigeon of partial information about the

torso The deletion method allows for the removal ofthe legs without influencing the visibility of the torsoFinally this also allowed us to delete additionalportions of the models (head torso) that would nothave been appropriate without putting an occluder inunnatural and unusual positions (although such digitaldeletions could also be considered unnatural)

Thus testing and comparing both occlusion anddeletion versions of these stimuli would best determinethe relative contributions of the different body parts tothe discrimination The resulting patterns of theoutcomes can provide insight into how differentportions of the models support the discrimination andtheir relative importance and how globalconfiguraland localfeatural information potentially work to-gether in mediating the discrimination of the modelsrsquoactions by the pigeons

Methods

The same three pigeons and apparatus were used asat the end of Experiment 2

Procedure

Occlusion test For the occlusion test a digital rock wasplaced in the scene with either the buck or dog model

Figure 6 Mean normalized peck rates for the three pigeons in Experiment 3 tested in different types of featured deletion conditions

Performance with various portions of the stimulus deleted Error bars indicate the standard error of each condition

This rock was placed in one of seven possible locations(see Figure 5 samples and Supplementary Movies 13ndash19) Horizontally the rock was positioned at themodelrsquos rear legs fore legs or extending across bothpairs of legs This horizontal rock position wasfactorially combined with the rockrsquos depth such thatthe rock was either in front of the model and occludingthe digital modelrsquos limbs or behind the digital modeland being occluded Finally to acclimate the pigeons tothe rock-containing displays and to evaluate theabsolute effect of the rockrsquos presence a beside conditionwas also used in which the rock was placed ahead of themodel so that at no point during locomotion did themodel reach the rock For the running buck model 24frames were used to comprise one cycle instead of the19 frames in baseline to create a slightly smoothermotion but this did not seem to affect performanceFor this test perspective was restricted to the lowclose side perspective

In the eight sessions prior to the tests the pigeonswere given training with the beside condition Thesewere added as eight trials to the baseline 72 ground-context trials two S one reinforced Sthorn and oneprobe Sthorn for both the buck and dog models

For test sessions the six rock-placement stimuli(behind vs occluding middot rear fore and both) for a givenmodel were conducted as nonreinforced probes Twelvetrials testing these conditions (six rock placements middottwo actions) the four trials of the beside condition (twoS one Sthorn one nonreinforced Sthorn) and the 72-trialbaseline trials comprised an 88-trial session Eight testsessions were conducted with the dog model and fourwere conducted with the buck model After every twoexperimental sessions one baseline session was given toreduce memorization of the experimental test stimuliProbe data in the rock-beside condition was notavailable for the first four sessions with the dog modelso the peck rates from the interleaved baseline sessionswere used for those data pointsDeletion test In this test five different part-deletedstimuli were tested to evaluate the pigeonsrsquo use of therear legs the fore legs all legs the torso midsectionand the head (see Figure 6 and Supplementary Movies20ndash24) These stimuli were generated by marking theindicated components of the buck and dog modelsinvisible to the rendering algorithm so that either theground or the background appeared where previouslythe body part was present These stimuli were thenrendered using the low close side perspective

Two different test sessions were composed for thisexperiment For each model one session tested thethree leg deletions (six test trials) and the second testedthe head and torso deletions (four test trials) All testtrials were conducted as nonreinforced probes andmixed in 72-trial baseline trials (78-trial or 76-trialsessions respectively) Four two-session blocks were

tested for each model (16 total sessions) with singlebaseline sessions separating each testing block

Results

Occlusion test Shown in Figure 5 are the results for thedifferent occlusion test conditions as a function ofmean normalized pecks to the display All three pigeonsshowed continued strong baseline discrimination of thetested models and perspective as measured by DR ts(2)frac14 169 p frac14 0003 dfrac14 97 The addition of anonoccluding rock beside the model resulted in a smalldecline in DR (fromMfrac14 87 SEfrac14 04 toMfrac14 80 SEfrac1412) but discrimination continued significantly abovechance t(2)frac14 44 p 0049 dfrac14 25

Averaging together the three occluding and threebehind rock conditions the three individual pigeonsshowed significantly above-chance discrimination ofthe transfer stimuli (G2 DRfrac14 89 S3 DRfrac14 65 Y4DRfrac14 68 ts(11) 39 ps 0002) Whether the rockwas placed in front of or behind the model made nodifference in the pigeonsrsquo reactions to the displays astheir discrimination was very similar across thismanipulation This equivalence was supported by arepeated-measures ANOVA (occluding vs behind middothorizontal rock position) using DR This analysisrevealed no main effect of the occluding versus behindfactor F(1 2) frac14 14 or its interaction with horizontalrock position F(2 4) 1 Horizontal rock position didshow a significant main effect F(2 4)frac14 77 pfrac14 0043g2pfrac14 79 which indicates that the location of the rockwith respect to the model affected discriminationIndividual differences among the pigeons however areessential for a complete understanding of this effect

To better evaluate how the position of the rockrelative to the body affected the discriminability of theactions the left half of Table 2 reports DR and itsanalyses for each condition as averaged across theocclusion and behind conditions because performancein those conditions seemed equivalent This analysisusing DR revealed that both G2 and S3 coulddiscriminate the actions regardless of the rockrsquoshorizontal position although the all legs condition wasnumerically worse Pigeon Y4 had difficulty when therear legs were unavailable as discrimination was lowerin both the rear legs and all legs conditions (in bothcases the statistical probabilities were marginal)Correspondingly this pigeon could easily discriminatethe actions when the rear legs were available (ie forecondition)Deletion test Figure 6 shows the normalized results forthe different deletion test conditions as a function ofmean peck rate The results reveal that selectivedeletion of the different parts of the models minimallyaffected discrimination Baseline DR with just the

models tested in the deletion tests continued to beexcellent t(2)frac14225 pfrac140004 dfrac14130 Discriminationwas well above chance for the part-deleted stimulioverall for each individual bird (G2 DRfrac14 89 S3 DRfrac14 067 Y4 DR frac14 088 ts(7) 50 ps 0002 ds 18) A one-way repeated-measures ANOVA (deletioncondition) on discrimination ratio revealed no differ-ences between the deletion conditions both whenconsidered across the three pigeons F(4 8) frac14 10 andwhen considered individually for each bird Fs(4 28) 14 This experiment however also benefits from closeattention to individual bird performance As Table 2shows although G2 and Y4 were able to discrimi-nate action across all deletion conditions S3 neededall the legs and the torso to accurately discriminate

Discussion

The results of these two analytic tests indicate thatno specific localized part of the model was the criticaldeterminant of the discrimination Although removingboth pairs of legs of a walking or running model waslikely most disruptive for all birds in the occlusion testthere was still enough information that two of threepigeons could discriminate between the actions Simi-larly as the different parts of the model were selectivelyremoved in the deletion test the remainder of themodel retained sufficient information to mediate thediscrimination Even when the legs were completely

removed most of the pigeons continued to recognizethe modelrsquos actions Thus it appears that as long as themajority of the model was available the pigeons wereable to continue identifying the modelsrsquo actions Theyclearly were not exclusively relying on a specificlocalized stimulus region or a particular body partsuch as the seemingly salient legs to discriminate theseactions The implication of this pattern is that thepigeons were globally or configurally evaluating themodelsrsquo bodily articulated movements over time

One curious outcome of the experiment was thatplacing the lsquolsquooccludingrsquorsquo rock behind the modelsrsquo legshad approximately the same impact on performance asplacing it in front The occlusion of the model wherefeatures were clearly hidden did not highly disrupt thediscrimination except perhaps when both pairs of legswere invisible Placing the rock behind both legs had asimilar impact on discrimination As mentioned in theIntroduction this is not the first time that this type oflsquolsquobehind-maskingrsquorsquo effect has been found for pigeons(DiPietro et al 2002 Koban amp Cook 2009 Lazarevaet al 2007) It is not clear why otherwise visibleinformation seems to be interfered with in suchconditions for pigeons although experience can helpreduce its effects (Lazareva et al 2007) Even themodelrsquos movements did not help to disambiguate thecritical shape and pose information from the mask Forreasons that are not entirely clear at the moment itappears that the introduction of novel backgroundregions creates problems for pigeons in decomposing or

Rock placement Deletions

Beside Rear Fore All Rear Fore All Torso Head

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

Table 2 Analysis of individual bird performance in Experiment 3 Notes The left half of the table shows performance with a digitalrock beside the acting model covering its rear legs fore legs or across both legs as averaged across the occludingbehindmanipulation The right half shows performance when the specified parts were deleted from the model All refers to the all legsoccluded or deleted conditions All bolded p values are significant after using a Holm-Bonferroni multiple test correction for each bird

segregating even familiar elements into separate partsBetter understanding how the edge and segregationprocesses by which pigeons visually decompose over-lapping objects is an important question for furtherexploration in the future as they may possibly divergefrom those in mammals

General discussion

These three experiments investigated for the firsttime the visual control of a locomotive actiondiscrimination by pigeons Testing different feature-altered digital models demonstrating each action theexperiments provide important new information aboutavian action recognition and motion processingExperiment 1rsquos two tests revealed that lsquolsquobiologicalmotionrsquorsquo stimuli (ie biomechanically coordinatedmoving dots) could not support discriminative transferfollowing extensive training with complete modelsExperiment 2 revealed that connected edge informationand interior shading were likely critical elements to thediscrimination as both silhouette and contour-onlystimuli were sufficient to support an above-chance butreduced level of transfer Experiment 3 suggested thatmotion information from both the body and legs wereinvolved in the pigeonsrsquo determination of the entiremodelrsquos actions Together this evidence indicates thepigeons are likely using the motion information derivedfrom the entire model to discriminate these walking andrunning actions As a consequence they support ahypothesis that pigeons can extract and classify theglobally organized locomotive actions of nonaviananimal models

This outcome is consistent with the different set offeature tests in the earlier results (Asen amp Cook 2012)That study had manipulated presentation rate andeliminated local rate of motion as a potential featureindicating that the pigeons were not simply looking forhow fast parts of the displays moved As found here aswell inversion of the displays disrupted discriminationSuch inversions retain a vast majority of local featuresand Experiment 3 provided evidence that the discrim-ination is not solely a product of attention to the legmotions which are most disrupted by inversion (cfHirai et al 2011) When combined with the presentresults it strongly suggests the pigeons were using theglobal configuration of the moving model instead ofrelying on local features If so these outcomes suggestthat the pigeons had learned to recognize sequences ofspatially oriented poses or that the relative configura-tions of body parts within the model are the bases ofthis locomotion discrimination (eg Singer amp Shein-berg 2010) The results of Experiments 2 and 3 seemconsistent with this hypothesis as different parts of the

modelrsquos body seemed sufficient to mediate the dis-crimination in the absence of specific parts Althoughthe relative motion and positioning of the legs likelycontributed significantly to the discrimination (it islocomotion after all) the pigeons were still able toidentify the modelrsquos actions when the legs were entirelyeliminated in Experiment 3 This indicates that thebodyrsquos movement also carries useful informationRecognizing such pose sequences seems to require theencoding of the relative global positions and motions ofmultiple body parts This type of configural encodingwas likely facilitated by our use of numerous modelsmultiple orientations and camera distances duringtraining Such extensive stimulus variability likelyencourages the pigeons to use generalized globalinformation from the modelrsquos movements instead oflocalized parts or specific locations to classify this largenumber of possible displays

If the pigeons attended to and integrated globalmovement information from across the completemodel there were some limitations in its application Inparticular this moving form information may need tobe bounded by a contour or filled in This is suggestedby the complete failure of the PLD stimuli to supportthis action discrimination Despite several manipula-tions designed to enhance the perceptual grouping ofthese coordinated but disconnected elements thepigeons failed to see any actions in this type of displayThe pigeonsrsquo failure to see lsquolsquobiological motionrsquorsquo couldrepresent either a perceptual or cognitive deficit Onepossibility is that pigeons have visual or perceptuallimitations on integrating unconnected parts into acoordinated whole Although pigeons can integrateglobal visual information (Cook 1992 2001 Cook etal 2005 Troje amp Aust 2013) there are also numerousinstances of pigeons showing difficulties in groupingseparated elements (Aust amp Huber 2006 Kelly et al2001 Sekuler et al 1996) exhibiting greater attentionto isolated local elements relative to the larger globalinformation (Cavoto amp Cook 2001 Lea et al 2006)and attending to separated information (M F BrownCook Lamb amp Riley 1984 Cook Riley amp Brown1992) The latter difficulties suggest that perceptuallyintegrating separated information is not necessarilyalways easy for them A second possibility is thatpigeons do not retain or use the same kind of higher-order cognitive models for action as humans might doIt is the possession of these top-down expectations andattention for actions that may allow humans to readilysee biological motion in such impoverished stimuli(Dittrich 1999 Shiffrar Lichtey amp Chatterjee 1997Thornton Rensink amp Shiffrar 2002) It is not clearwhat if any top-down expectations the pigeons mayhave had here Finally because of the considerabledifference in the visual appearance of the completemodels and the PLD stimuli a third possibility is that

the pigeons simply did not associate the PLD displayswith their previously reinforced discrimination and theresulting generalization decrement limits their transferReconciling the conditions that may facilitate pigeonsrsquoglobal representation of actions with PLD stimuli is animportant goal for future research

With the discovery of mirror neurons in monkeysthere has been renewed interest in motor-based theoriesof human action embodied cognition languageintentionality and social cognition (Engel et al 2013Gallese 2007 Grafton 2009 Iacoboni 2009 Jean-nerod 2001 Rizzolatti et al 2001 Wilson amp Knoblich2005) The current study of action discrimination by anonhuman animal adds significantly to the ongoingdebate regarding the nature and mechanisms of humanaction recognition and the role of motor simulation(Decety amp Grezes 1999 Heyes 2010 Hickok 2009)One collective theme of many proposals is that actionexecution and action observation are commonly codedsometimes suggested to be mediated by an independentspecies-specific action network (Grafton 2009 Jean-nerod 2001) Given that the motor systems forlocomotion and flying in pigeons share little in commonwith the different quadruped motions tested here ourresults carry the implication that actions can bediscriminated without simple embodiment within theobserver and without large computational and linguis-tic capacities Thus the current results seem to runcounter to a major prediction of theories that assumean overlap between action execution and actionperception Whatever species-specific action system oreven mirror-like neurons (Prather Peters Nowicki ampMooney 2008) birds may have it is likely not evolvedfor the present discrimination or the models tested Theresults instead suggest that the visual processesgenerally available for motion perception are likelysufficient to extract and recognize complex sequen-tially moving forms without requiring activation of orsimulation by the motor system Such a conclusion isconsistent with recent human findings regarding thediscrimination of biologically consistent and artificialtrajectories (Jastorff Kourtzi amp Giese 2006) andrelated criticisms of such motor-based theories (Hick-ok 2009) Animals regularly need to recognize andreact to the behaviors of a wide variety of species withwhich they may share few motor programs Whileembodied cognition makes good evolutionary sensewhen thinking about the origins of cognition generallymaking action recognition specifically dependent onyour speciesrsquo motor representations would preventeffective recognition of nonconspecific behavior

Examinations across different animal species willadd considerably to our understanding of mechanismsand evolution of behavioral recognition and advancethe development of an expanded comparative science ofvisual cognition With increasing success studies have

suggested that pigeons are able to form motion-basedaction categories (Asen amp Cook 2012 Cook Beale ampKoban 2011 Cook Shaw amp Blaisdell 2001 Dittrichet al 1998 Mui et al 2007) despite a size-limitednervous system The stimuli in this experiment focusedon locomotor categories because they are likely salientand tractable natural categories These actions are alsosimple periodic and repetitive In nature howeverthere are many examples of temporally extendedcomplex action series that comprise single behaviorssuch as grooming or courting (Shimizu 1998) Thecapacity to discriminate between such complex andnonrepetitive behaviors is clearly an important exten-sion of the present research requiring further investi-gation

Keywords pigeon action recognition biologicalmotion occlusion global perception

Acknowledgments

This research was supported by a grant from theNational Eye Institute (RO1-EY022655) E-mailRobertCooktuftsedu Home Pagewwwpigeonpsytuftsedu

Commercial relationships noneCorresponding author Muhammad A J QadriEmail MuhammadQadrituftseduAddress Department of Psychology Tufts UniversityMedford MA USA

References

Aggarwal J K amp Cai Q (1999) Human motionanalysis A review Computer Vision and ImageUnderstanding 73 428ndash440

Arbib M A (2005) From monkey-like actionrecognition to human language An evolutionaryframework for neurolinguistics Behavioral andBrain Sciences 28 105ndash124

Asen Y L amp Cook R G (2012) Discrimination andcategorization of actions by pigeons PsychologicalScience 23 617ndash624 doi1011770956797611433333

Aust U amp Huber L (2006) Does the use of naturalstimuli facilitate amodal completion in pigeonsPerception 35 333ndash349

Beintema J A amp Lappe M (2002) Perception ofbiological motion without local image motionProceedings of the National Academy of Sciences

USA 99(8) 5661ndash5663 doi101073pnas082483699

Blake R (1993) Cats perceive biological motionPsychological Science 4 54ndash57

Blake R amp Shiffrar M (2007) Perception of humanmotion Annual Review of Psychology 58 47ndash73

Bobick A F amp Davis J W (2001) The recognitionof human movement using temporal templatesPattern Analysis and Machine Intelligence IEEETransactions on 23(3) 257ndash267

Brown J Kaplan G Rogers L J amp VallortigaraG (2010) Perception of biological motion incommon marmosets (Callithrix jacchus) By femalesonly Animal Cognition 13(3) 555ndash564 doi101007s10071-009-0306-0

Brown M F Cook R G Lamb M R amp Riley DA (1984) The relation between response andattentional shifts in pigeon compound matching-to-sample performance Animal Learning amp Behavior12 41ndash49 doi103758BF03199811

Buccino G Binkofski F amp Riggio L (2004) Themirror neuron system and action recognition Brainand Language 89(2) 370ndash376

Buccino G Lui F Canessa N Patteri I Lagravi-nese G Benuzzi F Rizzolatti G (2004)Neural circuits involved in the recognition ofactions performed by nonconspecifics An fMRIstudy Journal of Cognitive Neuroscience 16(1)114ndash126

Byrne R W amp Russon A E (1998) Learning byimitation A hierarchical approach Behavioral andBrain Sciences 21 667ndash684

Cabe P A amp Healey M L (1979) Figure-background color differences and transfer ofdiscrimination from objects to line drawings withpigeons Bulletin of the Psychonomic Society 13(3)124ndash126

Cavoto K K amp Cook R G (2001) Cognitiveprecedence for local information in hierarchicalstimulus processing by pigeons Journal of Exper-imental Psychology Animal Behavior Processes27(1) 3ndash16 doi1010370097-74032713

Cook R G (1992) Dimensional organization andtexture-discrimination in pigeons Journal of Ex-perimental Psychology Animal Behavior Processes18(4) 354ndash363

Cook R G (2001) Hierarchical stimulus processingby pigeons In R G Cook (Ed) Avian visualcognition Available at wwwpigeonpsytuftseduavccook

Cook R G Beale K amp Koban A C (2011)Velocity-based motion categorization by pigeons

Journal of Experimental Psychology Animal Be-havior Processes 37 175ndash188 doi101037a0022105

Cook R G Goto K amp Brooks D I (2005) Aviandetection and identification of perceptual organi-zation in random noise Behavioural Processes69(1) 79ndash95 doi101016jbeproc200501006

Cook R G amp Katz J S (1999) Dynamic objectperception by pigeons Journal of ExperimentalPsychology Animal Behavior Processes 25(2) 194ndash210 doi1010370098-7403252194

Cook R G Riley D A amp Brown M F (1992)Spatial and configural factors in compound stim-ulus-processing by pigeons Animal Learning ampBehavior 20(1) 41ndash55

Cook R G amp Roberts S (2007) The role of videocoherence on object-based motion discriminationsby pigeons Journal of Experimental PsychologyAnimal Behavior Processes 33(3) 287ndash298 doi1010370097-7403333287

Cook R G Shaw R amp Blaisdell A P (2001)Dynamic object perception by pigeons Discrimi-nation of action in video presentations AnimalCognition 4 137ndash146 doi101007s100710100097

Cook R G Wright A A amp Drachman E E (2012)Categorization of birds mammals and chimeras bypigeons Behavioural Processes 93 98ndash110 doi101016jbeproc201211006

Decety J amp Grezes J (1999) Neural mechanismssubserving the perception of human actions Trendsin Cognitive Sciences 3(5) 172ndash178

di Pellegrino G Fadiga L Fogassi L Gallese V ampRizzolatti G (1992) Understanding motor eventsA neurophysiological study Experimental BrainResearch 91(1) 176ndash180

DiPietro N T Wasserman E A amp Young M E(2002) Effects of occlusion on pigeonsrsquo visualobject recognition Perception 31 1299ndash1312

Dittrich W H (1993) Action categories and theperception of biological motion Perception 22 15ndash22

Dittrich W H (1999) Seeing biological motion - Isthere a role for cognitive strategies In A BraffortR Gherbi S Gibet J Richardson amp D Teil(Eds) Gesture-based communication in human-computer interaction (Vol 1739) (pp 3ndash22) BerlinGermany Springer

Dittrich W H amp Lea S E G (1993) Motion as anatural category for pigeons Generalization and afeature-positive effect Journal of the ExperimentalAnalysis of Behavior 59 115ndash129 doi101901jeab199359-115

Dittrich W H Lea S E G Barrett J amp Gurr P R(1998) Categorization of natural movements bypigeons Visual concept discrimination and bio-logical motion Journal of the Experimental Anal-ysis of Behavior 70 281ndash299

Engel A K Maye A Kurthen M amp Konig P(2013) Wherersquos the action The pragmatic turn incognitive science Trends in Cognitive Sciences17(5) 202ndash209 doi101016jtics201303006

Fernandez-Juricic E Erichsen J T amp Kacelnik A(2004) Visual perception and social foraging inbirds Trends in Ecology amp Evolution 19 25ndash31

Gallese V (2007) Before and below lsquotheory of mindrsquoEmbodied simulation and the neural correlates ofsocial cognition Philosophical Transactions of theRoyal Society B Biological Sciences 362(1480)659ndash669

Giese M A amp Poggio T (2003) Neural mechanismsfor the recognition of biological movementsNature Reviews Neuroscience 4 179ndash192

Grafton S T (2009) Embodied cognition and thesimulation of action to understand others Annalsof the New York Academy of Sciences 1156(1) 97ndash117

Heyes C (2010) Mesmerising mirror neurons Neuro-Image 51(2) 789ndash791

Hickok G (2009) Eight problems for the mirrorneuron theory of action understanding in monkeysand humans Journal of Cognitive Neuroscience21(7) 1229ndash1243

Hirai M Chang D H F Saunders D R amp TrojeN F (2011) Body configuration modulates theusage of local cues to direction in biological-motionperception Psychological Science 22(12) 1543ndash1549 doi1011770956797611417257

Huber L Aust U Michelbach G Olzant SLoidolt M amp Nowotny R (1999) Limits onsymmetry conceptualization in pigeons The Quar-terly Journal of Experimental Psychology 52B 351ndash379

Iacoboni M (2009) Imitation empathy and mirrorneurons Annual Review of Psychology 60 653ndash670

Jastorff J Kourtzi Z amp Giese M A (2006)Learning to discriminate complex movementsBiological versus artificial trajectories Journal ofVision 6(8)3 791ndash804 httpwwwjournalofvisionorgcontent683 doi101167683 [PubMed] [Article]

Jeannerod M (2001) Neural simulation of action Aunifying mechanism for motor cognition Neuro-Image 14(1) S103ndashS109

Jitsumori M Natori M amp Okuyama K (1999)Recognition of moving video images of conspecificsby pigeons Effects of individuals static anddynamic motion cues and movement AnimalLearning amp Behavior 27 303ndash315

Johansson G (1973) Visual perception of biologicalmotion and a model of its analysis Perception andPsychophysics 14 201ndash211

Kaiser M D amp Shiffrar M (2009) The visualperception of motion by observers with autismspectrum disorders A review and synthesis Psy-chonomic Bulletin amp Review 16(5) 761ndash777 doi103758pbr165761

Kelly D M Bischof W F Wong-Wylie D R ampSpetch M L (2001) Detection of Glass patternsby pigeons and humans Implications for differ-ences in higher-level processing PsychologicalScience 12(4) 338ndash342

Koban A C amp Cook R G (2009) Rotational objectdiscrimination by pigeons Journal of ExperimentalPsychology Animal Behavior Processes 35 250ndash265 doi101037a0013874

Lazareva O F Wasserman E A amp Biederman I(2007) Pigeonsrsquo recognition of partially occludedobjects depends on specific training experiencePerception 36(1) 33ndash48

Lea S E G Goto K Osthaus B amp Ryan C M E(2006) The logic of the stimulus Animal Cognition9(4) 247ndash256 doi101007s10071-006-0038-3

Malt B C Gennari S Imai M Ameel E TsudaN amp Majid A (2008) Talking about walkingBiomechanics and the language of locomotionPsychological Science 19(3) 232ndash240

Mui R Haselgrove M McGregor A Futter JHeyes C amp Pearce J M (2007) The discrimi-nation of natural movement by budgerigars (Me-lopsittacus undulates) and pigeons (Columba livia)Journal of Experimental Psychology Animal Be-havior Processes 33 371ndash380

Oram M amp Perrett D (1994) Responses of anteriorsuperior temporal polysensory (STPa) neurons tolsquolsquobiological motionrsquorsquo stimuli Journal of CognitiveNeuroscience 6(2) 99ndash116

Parron C Deruelle C amp Fagot J (2007) Processingof biological motion point-light displays by ba-boons (Papio papio) Journal of ExperimentalPsychology Animal Behavior Processes 33 381ndash391

Peissig J J Young M E Wasserman E A ampBiederman I (2005) The role of edges in objectrecognition by pigeons Perception 34 1353ndash1374

Polana R amp Nelson R (1997) Detection and

recognition of periodic nonrigid motion Interna-tional Journal of Computer Vision 23(3) 261ndash282

Poppe R (2010) A survey on vision-based humanaction recognition Image and Vision Computing28(6) 976ndash990

Prather J F Peters S Nowicki S amp Mooney R(2008) Precise auditory-vocal mirroring in neuronsfor learned vocal communication Nature451(7176) 305ndash310 doi101038nature06492

Puce A amp Perrett D (2003) Electrophysiology andbrain imaging of biological motion PhilosophicalTransactions of the Royal Society of London SeriesB Biological Sciences 358(1431) 435ndash445

Regolin L Tommasi L amp Vallortigara G (2000)Visual perception of biological motion in newlyhatched chicks as revealed by an imprintingprocedure Animal Cognition 3(1) 53ndash60

Rizzolatti G Fogassi L amp Gallese V (2001)Neurophysiological mechanisms underlying theunderstanding and imitation of action NatureReviews Neuroscience 2 661ndash670

Schindler K amp Van Gool L (2008) Action snippetsHow many frames does human action recognitionrequire Proceedings of the Conference on ComputerVision and Pattern Recognition (pp 1ndash8)

Sekuler A B Lee J A J amp Shettleworth S J(1996) Pigeons do not complete partly occludedfigures Perception 25 1109ndash1120

Shiffrar M Lichtey L amp Chatterjee S H (1997)The perception of biological motion across aper-tures Perception amp Psychophysics 59(1) 51ndash59

Shimizu T (1998) Conspecific recognition in pigeons(Columba livia) using dynamic video imagesBehaviour 135 43ndash53

Singer J M amp Sheinberg D L (2010) Temporal

cortex neurons encode articulated actions as slowsequences of integrated poses Journal of Neurosci-ence 30 3133ndash3145

Thirkettle M Benton C P amp Scott-Samuel N E(2009) Contributions of form motion and task tobiological motion perception Journal of Vision9(3)28 1ndash11 httpwwwjournalofvisionorgcontent9328 doi1011679328 [PubMed][Article]

Thornton I M Rensink R A amp Shiffrar M (2002)Active versus passive processing of biologicalmotion Perception 31(7) 837ndash853

Tomonaga M (2001) Visual search for biologicalmotion patterns in chimpanzees (Pan troglodytes)Psychologia An International Journal of Psychologyin the Orient 44 46ndash59

Troje N F amp Aust U (2013) What do you meanwith lsquolsquodirectionrsquorsquo Local and global cues to biolog-ical motion perception in pigeons Vision Research79 47ndash55 doi101016jvisres201301002

Vallortigara G Regolin L amp Marconato F (2005)Visually inexperienced chicks exhibit spontaneouspreference for biological motion patterns PLoSBiology 3 1312ndash1316

Wang L Hu W amp Tan T (2003) Recentdevelopments in human motion analysis PatternRecognition 36(3) 585ndash601

Wilson M amp Knoblich G (2005) The case for motorinvolvement in perceiving conspecifics Psycholog-ical Bulletin 131(3) 460ndash473

Young M E Peissig J J Wasserman E A ampBiederman I (2001) Discrimination of geons bypigeons The effects of variations in surfacedepiction Animal Learning amp Behavior 29(2) 97ndash106

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

nition by nonhuman animals has progressed muchmore slowly because of the difficulty of controlling andusing lsquolsquobehaviorrsquorsquo in experimentally analytic situationsAnimals just donrsquot take direction well Digital softwareused to create animated displays of behavior howeverholds considerable promise for moving beyond thisproblem

Recently we successfully taught pigeons to discrim-inate and group the walking and running actions ofeight different digital animals using life-like articulat-ed animated models in a gono-go task (Asen amp Cook2012) Because these locomotor activities are likelysalient natural action categories (Malt et al 2008) theyprovided a good starting place for building on the priorresults of video-based action recognition (Dittrich ampLea 1993 Dittrich Lea Barrett amp Gurr 1998Jitsumori Natori amp Okuyama 1999) In that studyeach digital animal model ran or walked in place on atextured background (see examples in Figure 1) Toencourage action categorization the digital modelswere rendered from 12 different camera perspectives(combinations of elevation azimuth and distance) Itwas found that (a) this type of action discriminationwas easily acquired (b) it showed significant transfer tonovel species moving in biologically appropriate butdistinct ways (c) it exhibited viewpoint invariance overcamera distance elevation and perspective (d) it didnot vary substantially with variations in presentationspeed (e) and it showed selective interference with theinversion of the video or the randomization of itssequential frames The results seemed most consistentwith the hypothesis that the pigeons learned actioncategories for the different behaviors as a series of

sequenced poses Given that pigeon locomotion likelydoes not share motor representations in common withthe different quadruped actions that they were dis-criminating these results suggest that behaviors cansometimes be visually discriminated without theirembodiment in the observer

Computational models have explored the problemof human behavior recognition based exclusively onvisual information for a variety of functions (Aggar-wal amp Cai 1999 Poppe 2010 Wang Hu amp Tan2003) One computer vision approach focuses on thehigher-level global or configural organization amongdifferent body parts to recognize action The repre-sentation used in these theories often involveshierarchical geometry-based configural models cod-ing the relative motion of body limbs and joints(Aggarwal amp Cai 1999) A second approach codesnonconfigural and often nonparametric representa-tions to sufficiently discriminate among behaviorsThese theories vary in many ways including how andwhat information is encoded from global represen-tations such as space-time volumes or integratedsilhouettes to more localized features such as opticflow or periodic motion trajectories (eg Bobick ampDavis 2001 Polana amp Nelson 1997 Schindler amp VanGool 2008) One nonconfigural account of the actionsin Figure 1 for example might isolate the localizedmovement of the five different points traced in eachexample Given any one of these paths but especiallythose of the feet it would be possible to determine ifthe model were running or walking without processingthe entire figure Some models of human biologicalmotion perception utilize both configuraltop-down

Figure 1 Example of one of the eight animal models used in these experiments to exemplify the actions of walking and running It is

shown as rendered from a low close side perspective Superimposed on the displays are the different motion paths of five body

parts (nose neck junction tail junction fore right foot and rear right foot) These paths were not present in the stimuli tested with

the pigeons

Experiment 1

Methods

Animals

Apparatus

Procedure

Results

Discussion

Condition Dot size

Close Far

Sthorn S Sthorn S

Baseline 108 013 090 018

2middot 004 004 009 006

3middot 013 010 012 005

2middot 008 005 009 008

3middot 014 008 009 010

2middot 006 010 011 006

3middot 010 013 008 010

Experiment 2

Methods

Procedure

Results

Discussion

Experiment 3

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Experiment 1

Methods

Animals

Apparatus

Procedure

Results

Discussion

Condition Dot size

Close Far

Sthorn S Sthorn S

Baseline 108 013 090 018

2middot 004 004 009 006

3middot 013 010 012 005

2middot 008 005 009 008

3middot 014 008 009 010

2middot 006 010 011 006

3middot 010 013 008 010

Experiment 2

Methods

Procedure

Results

Discussion

Experiment 3

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Methods

Animals

Apparatus

Procedure

Results

Discussion

Condition Dot size

Close Far

Sthorn S Sthorn S

Baseline 108 013 090 018

2middot 004 004 009 006

3middot 013 010 012 005

2middot 008 005 009 008

3middot 014 008 009 010

2middot 006 010 011 006

3middot 010 013 008 010

Experiment 2

Methods

Procedure

Results

Discussion

Experiment 3

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Results

Discussion

Condition Dot size

Close Far

Sthorn S Sthorn S

Baseline 108 013 090 018

2middot 004 004 009 006

3middot 013 010 012 005

2middot 008 005 009 008

3middot 014 008 009 010

2middot 006 010 011 006

3middot 010 013 008 010

Experiment 2

Methods

Procedure

Results

Discussion

Experiment 3

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Results

Discussion

Condition Dot size

Close Far

Sthorn S Sthorn S

Baseline 108 013 090 018

2middot 004 004 009 006

3middot 013 010 012 005

2middot 008 005 009 008

3middot 014 008 009 010

2middot 006 010 011 006

3middot 010 013 008 010

Experiment 2

Methods

Procedure

Results

Discussion

Experiment 3

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Discussion

Condition Dot size

Close Far

Sthorn S Sthorn S

Baseline 108 013 090 018

2middot 004 004 009 006

3middot 013 010 012 005

2middot 008 005 009 008

3middot 014 008 009 010

2middot 006 010 011 006

3middot 010 013 008 010

Experiment 2

Methods

Procedure

Results

Discussion

Experiment 3

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Experiment 2

Methods

Procedure

Results

Discussion

Experiment 3

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Methods

Procedure

Results

Discussion

Experiment 3

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Results

Discussion

Experiment 3

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Experiment 3

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Methods

Procedure

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Results

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Discussion

DR 91 91 93 87 89 92 89 88 89

t(11) 326 42 373 158

t(7) 193 163 158 72 7

p 0001 0001 0001 0001 0001 0001 0001 0001 0001

d 94 121 108 46 68 58 56 26 25

DR 67 66 71 59 67 61 63 69 79

t(11) 39 45 73 27

t(7) 26 15 17 26 69

p 0002 0001 0001 0022 0038 0168 0141 0035 0001

d 11 13 21 08 09 05 06 09 24

DR 83 67 83 67 88 96 79 89 90

t(11) 59 2 72 21

t(7) 66 32 32 83 133

p 0001 0074 0001 0058 0001 0001 0025 0001 0001

d 17 06 21 06 23 113 11 29 47

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

General discussion

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Acknowledgments

References

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Introduction

Experiment 1

Experiment 2

Experiment 3

General discussion

Aggarwal1

Arbib1

Beintema1

Blake1

Blake2

Bobick1

Brown1

Brown2

Buccino1

Buccino2

Byrne1

Cavoto1

Decety1

diPellegrino1

DiPietro1

Dittrich1

Dittrich2

Dittrich3

Dittrich4

Engel1

FernandezJuricic1

Gallese1

Giese1

Grafton1

Heyes1

Hickok1

Hirai1

Huber1

Iacoboni1

Jastorff1

Jeannerod1

Jitsumori1

Johansson1

Kaiser1

Kelly1

Koban1

Lazareva1

Parron1

Peissig1

Polana1

Poppe1

Prather1

Regolin1

Rizzolatti1

Schindler1

Sekuler1

Shiffrar1

Shimizu1

Singer1

Thirkettle1

Thornton1

Tomonaga1

Troje1

Vallortigara1

Wilson1

Young1

Visual control of an action discrimination in pigeons · Visual control of an action discrimination in pigeons Muhammad A. J. Qadri $ Department of Psychology, Tufts University, Medford,

Documents