Beau­ty is no qual­i­ty in things them­selves: It exists mere­ly in the mind which con­tem­plates them; and each mind per­ceives a dif­fer­ent beau­ty (Hume, 1739/2012)

What does it mean if something is beautiful?

From the begin­ning, the devel­op­ment of tech­niques of cod­ing mean­ings into the frame­work of images has been accom­pa­nied by aes­thet­ic needs. Regard­less of whether or not we are aware of it, we eval­u­ate the objects we observe by cat­e­goris­ing them into those which we like and enjoy, or even desire, and those which we reject as val­ue­less and unsight­ly. Eval­u­a­tion of objects does not exclu­sive­ly refer to works of art, even though most peo­ple are con­vinced that this cat­e­go­ry of objects is par­tic­u­lar­ly sub­ject to aes­thet­ic assessment.

It is paint­ings, sculp­tures, archi­tec­ture or decor that are most often eval­u­at­ed as beau­ti­ful or ugly. The idea that we apply this cat­e­go­ry to objects of every­day use less fre­quent­ly is but an illu­sion. As Ernst Gom­brich (2005) points out, the essence of aes­thet­ic expe­ri­ence is by no means based on any def­i­n­i­tion of what is or should be treat­ed as Art, espe­cial­ly when writ­ten with a cap­i­tal “A”, but rather on the feel­ing that some­thing sim­ply looks “good”, regard­less of whether it hangs on a muse­um wall or serves as a dress accessory.

A sim­i­lar view is pre­sent­ed by Jean-Marie Scha­ef­fer (1999; 2000), empha­sis­ing the sub­jec­t’s deci­sive role in deter­min­ing the aes­thet­ic val­ue of an object. Sim­i­lar­ly to per­ceiv­ing a col­or or shape, the impres­sion of com­muning with a beau­ti­ful object is a sub­jec­tive expe­ri­ence rather than an innate trait of that object (Gołębiows­ka, 2005). Such a con­cep­tu­al­i­sa­tion of aes­thet­ic expe­ri­ence opens up an aes­thet­ics research field for a psy­chol­o­gist who requires oper­at­ing cat­e­gories that are seman­ti­cal­ly much sim­pler than those pro­posed by aes­thet­ics-relat­ed the­o­ries built at the inter­face of phi­los­o­phy, anthro­pol­o­gy and art history.

The intu­ition that some­thing looks good, bet­ter, or worse, is spon­ta­neous­ly ver­balised by means of expres­sions: “I like it,” “I don’t like it,” “I like it more” or “less”. It is of sig­nif­i­cant impor­tance for psy­chol­o­gist inter­est­ed in research­ing aes­thet­ic phe­nom­e­na because these expres­sions con­sti­tute use­ful behav­iour­al indi­ca­tors of the pecu­liar state of mind that we com­mon­ly refer to as aes­thet­ic judge­ment or, in Kan­t’s terms, judge­ment of taste (Kant, 1790/2004). Recog­nis­ing some­thing as pret­ty or good – with­in the mean­ing assigned to this word by Ernst Gom­brich (2005) – con­sti­tutes a con­ve­nient start­ing point to search for the deter­mi­nants of clas­si­fy­ing an object based on its aes­thet­ic values.

The need for aes­thet­ic eval­u­a­tion and com­mu­nion with beau­ty is also an ema­na­tion of bio­log­i­cal­ly deter­mined adap­ta­tion mech­a­nisms that allow one to sur­vive in a com­plex, often intim­i­dat­ing real­i­ty. It is a man­i­fes­ta­tion of the Freudi­an life dri­ve (Eros) as a counter-bal­ance to the destruc­tive death dri­ve (Tanatos) as the con­cept of being visu­al­ly “pret­ty” stands for numer­ous cat­e­gories which refer to var­i­ous sen­su­al modal­i­ties, such as “good”, “tasty”, “safe”, “sooth­ing”, “relax­ing”, “nice”, “seduc­tive” or “cheer­ful”. All of these con­cepts hold a def­i­nite­ly pos­i­tive con­no­ta­tion. They sig­ni­fy states of mind that we par­tic­u­lar­ly require to live, and there­fore they must undoubt­ed­ly have a sol­id neu­ro­bi­o­log­i­cal basis.

What is the nature of these sub­tle mech­a­nisms of expe­ri­ence and aes­thet­ic eval­u­a­tion of per­ceived images? What behav­iour­al (cog­ni­tive, emo­tion­al or moti­va­tion­al) dimen­sions can be used to describe them? Can they be cap­tured by means of neu­ro­phys­i­o­log­i­cal (ocu­lo­mo­tor, encephalo­graph­ic, elec­tromyo­graph­ic) indi­ca­tors and neu­roimag­ing? When it comes to con­tem­po­rary empir­i­cal sci­ences, includ­ing espe­cial­ly cog­ni­tive psy­chol­o­gy, neu­ro­bi­ol­o­gy and neu­ro­phys­i­ol­o­gy, these ques­tions have only begun to be posed. Most of them are not yet ful­ly-devel­oped deci­sive ques­tions that lie at the bases of sci­en­tif­ic hypothe­ses. They are rather explorato­ry in nature, test­ing the intu­itions derived from more or less spec­u­la­tive the­o­ries from the fields of aes­thet­ics, anthro­pol­o­gy, his­to­ry of art and philosophy.

Experience and aesthetic judgement

One of the most impor­tant, sub­jec­tive fac­tors con­di­tion­ing an aes­thet­ic judg­ment is aes­thet­ic expe­ri­ence. It is a reac­tion to the object of pre-reflec­tive and emo­tion­al per­cep­tion, which pre­cedes the reflec­tive aes­thet­ic judg­ment. Accord­ing to Jean-Marie Scha­ef­fer (2000), an aes­thet­ic judg­ment is an expres­sion of rec­on­cil­i­a­tion of val­ue with the sub­ject of aes­thet­ic experience.

Although the dis­tinc­tion between aes­thet­ic reac­tions to: sur­vival and judg­ment and the chrono­log­i­cal order of these reac­tions (first sur­vival and then judg­ment) are accept­ed by many aes­thet­ic the­o­rists (Gołaszews­ka, 1984; Tatarkiewicz, 2009), in research con­duct­ed on the basis of psy­chol­o­gy or neu­ro-aes­thet­ics they are rarely tak­en into account (Var­tan­ian and Goel, 2004a). In the vast major­i­ty of psy­cho­log­i­cal exper­i­ments, the task of their par­tic­i­pants is to assess the aes­thet­ic repro­duc­tion of works of art that are pre­sent­ed on a com­put­er mon­i­tor for a lim­it­ed time. Such a pro­ce­dure is more like test­ing the per­cep­tive­ness of dri­ving license stu­dents than con­tem­plat­ing works of art at the muse­um. The basic con­di­tion for an aes­thet­ic expe­ri­ence is the free time used for con­tact with the work.

Oculomotor indicators of mental processes

The pur­pose of research con­duct­ed in our lab­o­ra­to­ry is, among oth­ers, search­ing for ocu­lo­mo­tor cor­re­lates of expe­ri­ence and aes­thet­ic judg­ment. Eye move­ment reg­is­tra­tion seems to be an attrac­tive tool for exam­in­ing aes­thet­ic pref­er­ences because it pro­vides objec­tive indi­ca­tors of the strat­e­gy of view­ing of objects being eval­u­at­ed. After all, with our eyes closed, we can­not make an aes­thet­ic eval­u­a­tion of the image we have nev­er seen. In turn, with eyes open, we move them, focus­ing the visu­al axes on var­i­ous frag­ments of the viewed work and on this basis we eval­u­ate it.

Under­ly­ing the ocu­lo­mo­tor exam­i­na­tion is the assump­tion that the view­er most often looks at those parts of the image that he or she thinks about or which are of par­tic­u­lar impor­tance to him/her. Also, the time spent on view­ing of these frag­ments is an impor­tant indi­ca­tor of the inter­est and depth of pro­cess­ing of sen­so­ry data col­lect­ed dur­ing visu­al fix­a­tion. Gen­er­al­ly speak­ing, eye move­ment tra­jec­to­ry para­me­ters are inter­pret­ed as indi­ca­tors of men­tal process­es whose pur­pose is to inte­grate infor­ma­tion about the seen scene, as well as its eval­u­a­tion (Mas­saro, Savazzi, Di Dio, Freed­berg et al., 2012; Poole and Ball, 2005).

Pre­sent­ing in this chap­ter the results of the exper­i­ment on beau­ty per­cep­tion, I focused on analysing data con­cern­ing the time, num­ber, places of visu­al fix­a­tion on the paint­ings (and their frag­ments) pre­sent­ed to the sub­jects, length and speed of sac­cades, as well as the fre­quen­cy of sac­cades return­ing to places pre­vi­ous­ly viewed, i.e. the so-called regres­sion. These indi­ca­tors are divid­ed into two main groups: fix­a­tions and saccades.


Fix­a­tions can be described by three para­me­ters: time, num­ber and fre­quen­cy. The fix­a­tion time is divid­ed into fix­a­tion dura­tion total and fix­a­tion dura­tion average.

Fix­a­tion dura­tion total is about 90% of the total (glob­al) image view­ing time, so it is most often inter­pret­ed as a gen­er­al indi­ca­tor of inter­est in it or an indi­ca­tor of dif­fi­cul­ty in obtain­ing unam­bigu­ous per­cep­tu­al infor­ma­tion about it (Hauland, 2003; Latimer, 1988; Mel­lo-Thoms, Nodine and Kun­del, 2002). Fix­a­tion dura­tion total can refer to both the entire image space and its frag­ment. If it refers to a small part of an image then it is referred to as a gaze. Gaze is the most com­mon­ly used indi­ca­tor of visu­al atten­tion, pre­sent­ed graph­i­cal­ly on atten­tion­al maps or, in oth­er words, heat maps (Hoff­man, Grimes, Shon and Rao, 2007; Stell­mach, Nacke and Dachselt, 2011). These are the mul­ti-col­ored spots applied to the viewed image. I have used this way of pre­sent­ing data many times, illus­trat­ing some of the effects dis­cussed in this book.

In turn, the aver­age fix­a­tion time, which is the quo­tient of the fix­a­tion dura­tion total by their num­ber, is an indi­ca­tor of the increased involve­ment of visu­al atten­tion in the accu­rate explo­ration and inter­pre­ta­tion of the image or its frag­ment, as well as the depth of sen­so­ry data pro­cess­ing (Buswell, 1935; Duchows­ki, 2007; Gold­berg and Kot­val, 1999; Just and Car­pen­ter, 1976). Short­er aver­age fix­a­tion times are asso­ci­at­ed with more com­plex, more detailed images com­pared to less com­plex, less light­ed or blurred images (Duchows­ki, 2007; Mol­nar, 1981).

An inter­est­ing indi­ca­tor of image explo­ration is also the time after which the view­er first looks at the area, which for some rea­son is inter­est­ing for the researcher. It is the so-called time to first fix­a­tion on tar­get. The area where the observer’s gaze is direct­ed the fastest from the moment the expo­sure begins, most like­ly meets the cri­te­ria for the task of view­ing it. This eye move­ment is a man­i­fes­ta­tion of invol­un­tary atten­tion (Byrne, Ander­son, Dou­glass and Mates­sa, 1999).

The fix­a­tion count, like the fix­a­tion dura­tion total, is a gen­er­al indi­ca­tor of inter­est in an image. There is a high cor­re­la­tion between these two indi­ca­tors: the more fix­a­tion points are, the longer their dura­tion total is. The fix­a­tion count is an impor­tant indi­ca­tor of view­ing a paint­ing, espe­cial­ly when we focus not so much on its entire sur­face but on its var­i­ous parts.

Greater clus­ter of fix­a­tions on a small area of ​​a paint­ing, the so-called area of ​​inter­est (AOI) can mean either an increase in dif­fi­cul­ty in rec­og­niz­ing the objects depict­ed on it (Jacob and Karn, 2003), or the spe­cial impor­tance of this area for cap­tur­ing the sense of the entire scene depict­ed (Poole , Ball and Phillips, 2004). The small­er the area of ​​inter­est deter­mined by the fix­a­tion count and clus­ter of fix­a­tions, the more inquis­i­tive and tar­get­ed the search of this field is. On the con­trary, the more dif­fused the fix­a­tion points locat­ed in a small area, the less focused its view­ing (Cowen, Ball and Delin, 2002). 

The oppo­site of the fix­a­tion dura­tion aver­age is the fix­a­tion fre­quen­cy. It is the quo­tient of the fix­a­tion count (in rela­tion to the whole paint­ing or select­ed areas of inter­est) by their dura­tion total. A high­er fix­a­tion fre­quen­cy means a greater count of analysed fix­a­tion points per unit of time, e.g. per sec­ond. It can be an indi­ca­tor of emo­tion­al arousal result­ing from per­form­ing a task under time pres­sure or a strong need to iden­ti­fy dif­fi­cult to rec­og­nize things in the part of the pic­ture being viewed.


The sec­ond group of eye tra­jec­to­ry para­me­ters refers to sac­cades, i.e. shifts in the visu­al axis from one posi­tion to anoth­er. Sac­cade dura­tion total is about 10% of the total image view­ing time, while the sac­cade count is equal to the count of fix­a­tion points minus 1. Some­times sta­tis­ti­cal analy­ses use the sac­cade dura­tion aver­age, i.e. the quo­tient of the sac­cade dura­tion total to their count.

The dura­tion of indi­vid­ual sac­cades is an order of mag­ni­tude short­er than the fix­a­tion dura­tion and gen­er­al­ly do not dif­fer sig­nif­i­cant­ly from each oth­er. How­ev­er, if there are any dif­fer­ences found between them, they most often result from the dis­tance that divides the next two points that inter­sect the visu­al axis. It is the dis­tance between suc­ces­sive fix­a­tion points that is the most impor­tant char­ac­ter­is­tic of the sac­cade. It is called the sac­cade ampli­tude and most often the sac­cade ampli­tude aver­age is analysed. This is a scan­path strategy.

Wolf­gang H. Zange­meis­ter, Kei­th R. Sher­man and Lawrence Stark (1995) reg­is­tered the eye­ball move­ment of art experts and laypeo­ple while view­ing real­is­tic and abstract images. They divid­ed the sac­cades into glob­al (longer than 1.6°) and local (short­er than 1.6°). It turned out that experts were much more like­ly to use glob­al image search­ing strate­gies, while lay­men were more like­ly to use local strategies.

Both visu­al scene search strate­gies (glob­al and local) can be moti­vat­ed in a “bot­tom-up” or “top-down” man­ner (Mas­saro, Savazzi, Di Dio, Freed­berg, et. al., 2012). The bot­tom-up strat­e­gy of the per­cep­tu­al field search is based on struc­tur­al fea­tures of the image such as con­trast, col­or or com­po­si­tion. They attract the observer’s eye and focus his or her atten­tion. Lau­rent Itti and Christof Koch (2001) proved that “bot­tom — up” visu­al salience of var­i­ous image ele­ments can sig­nif­i­cant­ly con­tribute to form­ing the scan­path, name­ly the total length of all sac­cades (scan­path length), and, there­by, indi­rect­ly influ­ence their ampli­tude. In addi­tion, bot­tom-up moti­vat­ed strate­gies are more char­ac­ter­is­tic of the novices in the field the visu­al scene con­cerns than of the experts (Humprey and Under­wood, 2009).

On the oth­er head the top-down strat­e­gy of search­ing the field of per­cep­tion is to a greater extent con­trolled by knowl­edge and the atti­tude of the observ­er towards image explo­ration, result­ing form some ques­tion or aimed at ver­i­fy­ing a hypoth­e­sis.  Top-down strate­gies are more often applied by experts in the field the image per­tains to than by novices (Humprey and Under­wood, 2009). 

For the pur­pos­es of ocu­lo­graph­ic data analy­ses the fre­quen­cy of the so-called regres­sive sac­cades or, short­er, regres­sions, is tak­en into con­sid­er­a­tion.  Regres­sions are expressed by mul­ti­ple relo­ca­tion of the visu­al axes onto a pre­vi­ous­ly observed ele­ment of a visu­al scene (Sib­ert, Gok­turk and Lavine, 2000).  Their fre­quen­cy indi­cates the need for repeat­ed explo­ration of the area.  The rea­son for the regres­sion are most often dif­fi­cul­ties with read­ing the sense of the image con­tained in the ele­ment, due to its visu­al ambiguity. 

Regres­sion are espe­cial­ly observed while the exam­ined peo­ple are per­form­ing tasks which require sequen­tial relo­ca­tion of vision in one direc­tion, for exam­ple while read­ing.  Peo­ple who read poor­ly have a lot more regres­sive sac­cades than peo­ple who read very well (Rayn­er, 1975; Rayn­er and Pol­lat­sek, 1989; Rayn­er, Slat­tery and Belanger, 2010).

Research questions

After this brief dis­cus­sion of the para­me­ters of eye­ball move­ments, I wish to present the results of one of the tests con­duct­ed in our lab­o­ra­to­ry. Its aim was to deter­mine whether and what are the dif­fer­ences between the ocu­lo­mo­tor indi­ca­tors when view­ing images clas­si­fied by the sub­jects as beau­ti­ful and as non-beau­ti­ful (this does not mean ugly, but not belong­ing to the cat­e­go­ry of beau­ti­ful images). To a large extent, these stud­ies were explorato­ry in nature because no ocu­lo­graph­ic exper­i­ments had been con­duct­ed so far, in which the analy­sis of aes­thet­ic pref­er­ences was per­formed tak­ing into account the divi­sion into expe­ri­ence and aes­thet­ic judgement.

The moti­va­tion to under­take this study was also the search for an answer to the ques­tion whether the ocu­lo­mo­tor activ­i­ty accom­pa­ny­ing the aes­thet­ic expe­ri­ence (oper­a­tional­ly defined as free view­ing of images) dif­fers from the ocu­lo­mo­tor activ­i­ty when mak­ing aes­thet­ic judgments.

A few words on the method

The exper­i­ment referred to in the next chap­ter was con­duct­ed in an acousti­cal­ly iso­lat­ed stu­dio in the Psy­choneu­ro­phys­i­o­log­i­cal Lab­o­ra­to­ry by the Depart­ment of Exper­i­men­tal Psy­chol­o­gy of the Lublin Catholic Uni­ver­si­ty. SMI iView X Hi Speed ocu­lo­graph with 1250 Hz sam­pling fre­quen­cy, 0.5 mil­lisec­ond laten­cy and 0.01° res­o­lu­tion was used to mea­sure eye­ball move­ment. The stim­uli were dis­played on a 23″ Apple Cin­e­ma HD Dis­play screen (1920x1200 pix­els) at a dis­tance of 60 cen­time­tres from the eyes of the sub­jects (which cor­re­sponds to a 28.65 view angle). The sub­jects pro­vid­ed their respons­es using an Ergodex DX1 key­board with var­i­ous but­ton layouts.

The pro­ce­dures for dis­play­ing stim­uli and record­ing ocu­lo­mo­tor and behav­iour­al data were writ­ten by Paweł Augustynow­icz in E‑Prime v. 2.0. I car­ried out the analy­sis of behav­iour­al and ocu­lo­graph­ic data with OGAMA v. 4.0, BeGaze (for iView X SMI) and the STATISTICA v. 9.0 soft­ware pack­age, main­ly using vari­ance analy­sis. The lab­o­ra­to­ry exper­i­ments were super­vised by Bib­ian­na Bałaj. The research was anony­mous, and those who par­tic­i­pat­ed did not receive any addi­tion­al compensation.

The exper­i­ment con­sist­ed of two con­sec­u­tive parts. In the first one, the exam­ined per­sons were asked to see a set of sev­er­al dozen por­traits from the col­lec­tion of the Muse­um Palace at Wilanów. The paint­ings were exhib­it­ed in a ran­dom order, deter­mined indi­vid­u­al­ly for each respon­dent. The time dur­ing which the paint­ings could be viewed was unlim­it­ed. The respon­dent could look at each piece of art as long as they want­ed. To view the next paint­ing, they pressed the respec­tive key on the keyboard.

In order to stan­dard­ise the research sit­u­a­tion, before the first part of the exper­i­ment, the respon­dents were instruct­ed to look at the paint­ings in such a way that, after see­ing the entire col­lec­tion, they could point to those they con­sid­ered to be par­tic­u­lar­ly beau­ti­ful. Accord­ing to the instruc­tions, the exam­ined per­sons learned that the Muse­um Palace at Wilanów is prepar­ing for an inter­na­tion­al con­fer­ence devot­ed to the most beau­ti­ful por­traits from its own col­lec­tion. Accord­ing to the con­cept of the organ­is­ers, the con­fer­ence is to be accom­pa­nied by an exhi­bi­tion of paint­ings select­ed on the basis of the high­est aes­thet­ic val­ues. Not only art his­to­ri­ans and muse­um work­ers, but also a wide range of art view­ers were asked to point to these paint­ings. The aim of this study is to select 8–10 most beau­ti­ful pieces of art from the collection.

The unlim­it­ed time to view the paint­ings dur­ing the first part of the exper­i­ment and the instruc­tion to look at the paint­ings in terms of their aes­thet­ic val­ues cre­at­ed an oppor­tu­ni­ty for the respon­dents to spon­ta­neous­ly enjoy an aes­thet­ic expe­ri­ence. Not only did they not have to per­form tasks under time pres­sure, but what is more, they were con­vinced of the high val­ue of their own eval­u­a­tion, regard­less of any social norms, beliefs or trends that might influ­ence them.

In the sec­ond part of the exper­i­ment, the paint­ings that had already been viewed were pre­sent­ed again in ran­dom order. This time the respon­dents were asked to decide whether they would place the paint­ing on the list of the most beau­ti­ful ones or not. They sig­nalled their deci­sion by press­ing the appro­pri­ate key on the key­board. In this part, they direct­ly expressed their assess­ment of the piece of art in terms of aes­thet­ics. As in the pre­vi­ous stage, we did not lim­it the time spent on view­ing the paint­ing in this phase. Dur­ing both parts of all the exper­i­ments, we record­ed the eye­balls move­ment, reac­tion time and deci­sions of the exam­ined per­sons with regard to all the paint­ings being presented.


Why portraits?

Por­trait is one of the old­est forms of depict­ing peo­ple in paint­ings or in a form of sculp­tures (Freed­berg, 2005; Sufi, 2001). In per­cep­tu­al terms, rep­re­sen­ta­tion of a human being in a pic­ture, espe­cial­ly his or her face, has a sim­i­lar sta­tus as watch­ing it “live”. The human face is per­haps the strongest visu­al stim­u­lus that focus­es atten­tion on itself to an incom­pa­ra­bly greater degree than any oth­er ele­ment of the visu­al scene. This phe­nom­e­na was described by Guy Thomas Buswell in the 1930s, when he pre­sent­ed the results of ocu­log­ra­phy stud­ies in his clas­si­cal work: How peo­ple look at pic­tures (1935). The aim of his research was to dis­cov­er the reg­u­lar­i­ties gov­ern­ing the eye­ball move­ments when look­ing at pic­tures of dif­fer­ent con­tent. He called faces on the paint­ings as the prin­ci­pal cen­tres of interest.

The results of Buswell’s pio­neer­ing analy­ses were lat­er repeat­ed­ly con­firmed in stud­ies in which pic­tures depict­ing human faces were used as stim­u­lus mate­r­i­al (Land, Tatler, 2009; Yarbus, 1967). The result was always the same. Regard­less of whether the por­trayed per­sons were pret­ty or ugly, regard­less of paint­ing tech­nique used to illus­trate them, they always focused the audi­ence atten­tion first an to a great­est lev­el, com­pared with all the oth­er ele­ments of the pre­sent­ed scene.

Thus, ask­ing the ques­tion con­cern­ing how peo­ple look at pic­tures con­tain­ing the images of peo­ple, the fact whether their faces are pret­ty or ugly was not my con­cern. First of all, I want­ed to know whether dur­ing look­ing at pic­tures that are con­sid­ered to be beau­ti­ful as a whole, the exam­ined per­sons move their eyes in the same way as when look­ing at pic­tures that are rat­ed sig­nif­i­cant­ly low­er in terms of their aes­thet­ic val­ue. Of course I haven’t failed to check if the faces in the pic­tures that were eval­u­at­ed as beau­ti­ful are looked at in the same way as the faces in the pic­tures regard­ed as non-beau­ti­ful. It turned out that yes indeed, regard­less of the fact whether faces are beau­ti­ful or not — peo­ple devote equal­ly much atten­tion to por­trayed faces.

Oshin Var­tar­i­an and Vin­od Goel (2004b) found sig­nif­i­cant cor­re­la­tion between the time of view­ing pic­tures and aes­thet­ic pref­er­ence. Peo­ple look longer at images which are more beau­ti­ful. It is con­firmed also by the results of oth­er stud­ies (Hauland, 2003; Latimer, 1988; Mel­lo-Thoms, Nodine and Kun­del, 2002). There­fore, if the time devot­ed to look­ing at faces is sim­i­lar regard­less of whether they are ugly or pret­ty, to what else do they devote time while look­ing at pic­tures, which they claim are beau­ti­ful?  In these stud­ies, the issue that inter­est­ed me the most was the search for ocu­lo­mo­tor cor­re­lates between the aes­thet­ic expe­ri­ence and the aes­thet­ic judge­ment crown­ing it, while look­ing at whole pic­tures, an ele­ment of which were faces of por­trayed people.

Stimuli from the Wilanów collection and the research subjects

The Muse­um Palace at Wilanów gath­ers col­lec­tions of works of art relat­ed to place and per­sons who have shaped its image for more than three hun­dred years. Por­traits con­sti­tute a larg­er part of the col­lec­tion and they were also cho­sen as the stim­uli in these stud­ies.  Upon my request to the admin­is­tra­tion of the Muse­um I was giv­en a set of over 50 high-res­o­lu­tion paint­ing repro­duc­tions, which were select­ed by the art his­to­ri­ans work­ing in the muse­um as the most valu­able. These paint­ings were cre­at­ed in the 16th, 17th and 18th cen­tu­ry, and all of them depict­ed at least one person. 

After dis­card­ing a dozen or so paint­ings for tech­ni­cal rea­sons or due to the fact that they depict­ed famous peo­ple, for exam­ple king John III Sobies­ki, I final­ly decid­ed to use 32 repro­duc­tions.  These paint­ings were placed on grey boards and evened out on the basis of their height, regard­less of their actu­al dimen­sions (Fig. 162 A and B).

Fig­ure 162 A. Exam­ple board with Wilanów por­trait, used in research. Marceli Bac­cia­rell, Por­trait of Izabela Lubomirs­ka (1757).  Muse­um of King Jan III’s Palace at Wilanów, War­saw [139 x 101 cm]
Fig­ure 162 B. Exam­ple board with Wilanów por­trait, used in research. Hyacinthe Rigaud, Por­trait of Rigaud Kon­stan­ty Sobies­ki (1696). Muse­um of King Jan III’s Palace at Wilanów, War­saw [112.5 x 89 cm]

24 peo­ple (14 women and 10 men) aged 17 to 30 years (M = 24, SD = 4.19) with cor­rect visu­al acu­ity or the visu­al acu­ity cor­rect­ed to nor­mal were invit­ed to par­tic­i­pate in the study. None of them prac­ticed visu­al arts or had any edu­ca­tion in art or art his­to­ry. Their inter­est in paint­ing could be described as typ­i­cal in rela­tion to most muse­um visitors.

Beautiful and non-beautiful portraits

In accor­dance with the pro­ce­dure described in the intro­duc­tion to this chap­ter, at the sec­ond stage of this exper­i­ment the sub­jects chose the paint­ings that they con­sid­ered the most beau­ti­ful in the col­lec­tion they had viewed. They did not make that choice while view­ing them dur­ing the first part, but rather after becom­ing famil­iar with all of them. It allowed them to make an eval­u­a­tion while tak­ing into con­sid­er­a­tion the entire­ty of the exhi­bi­tion. The aes­thet­ic val­ue of each paint­ing was deter­mined by how often it was indi­cat­ed as beau­ti­ful by the sub­jects. There were no lim­i­ta­tions con­cern­ing the num­ber of paint­ings that each sub­ject could choose.

Based on the aes­thet­ic eval­u­a­tions record­ed dur­ing the sec­ond part of the exper­i­ment, all paint­ings were divid­ed into four cat­e­gories (with 8 paint­ings in each cat­e­go­ry). Cat­e­go­ry A includ­ed those paint­ings which were most fre­quent­ly indi­cat­ed by the sub­jects as the most beau­ti­ful. Cat­e­gories B and C includ­ed those which were less often select­ed for their aes­thet­ic val­ue, while Cat­e­go­ry D con­sist­ed of those paint­ings which were the least-often deemed beautiful.

In Fig. 163 A and B and Fig. 164 A and B I present the paint­ings clas­si­fied only in two extreme cat­e­gories: A (beau­ti­ful) and D (non-beau­ti­ful). Paint­ings clas­si­fied in cat­e­go­ry D can­not be con­sid­ered ugly or com­plete­ly devoid of aes­thet­ic val­ue. There­fore, I use the term “non-beau­ti­ful” to describe them, not “ugly” or “devoid of aes­thet­ic val­ue”. The num­bers below each image reveal its aes­thet­ic val­ue out of all thir­ty-two paint­ings. Num­ber (1) refers to the paint­ing that was most often indi­cat­ed as beau­ti­ful by the respon­dents, and num­ber (32) — least frequently.

Fig­ure 163 A. Aes­thet­ic cat­e­go­ry: A. (1) Angel­i­ca Kauff­mann, Por­trait of Krysty­na Potoc­ka (1783/ 1784) [155 x 113 cm], (2) Pom­peo Giro­lamo Batoni, Apol­lo and two Mus­es (after 1741) [122 x 90 cm], (3) Tin­toret­to Fam­i­ly Work­shop (?), Apoth­e­o­sis of a Venet­ian Sen­a­tor (the end of 16th cen­tu­ry) [175,6 x 174,7 cm], (4) Jan Gos­saert called Mabuse, Madon­na and the Child (around 1533?) [94 x 68 cm]. Muse­um of King Jan III’s Palace at Wilanów, Warsaw
Fig­ure 163 B. Aes­thet­ic cat­e­go­ry: A. (5) Jacques Loius David, [Eques­tri­an] Por­trait of Count Stanis­las Potoc­ki (1781) [304 x 218 cm], (6) Domeni­co Zampieri called Domenichi­no, Guardian Angel [132 x 100 cm], (7) Pom­peo Giro­lamo Batoni, Por­trait of Izabela Potoc­ka as Poli­hym­nia (1780) [65 x 53 cm], (8) Anton Graff, Por­trait of Stanis­laus Kost­ka (1785) [68 x 55 cm]. Muse­um of King Jan III’s Palace at Wilanów, Warsaw
Fig­ure 164 A. Aes­thet­ic cat­e­go­ry: D. (25) Jerzy Eleuter Szy­monow­icz-Siemigi­nows­ki, Por­trait of Queen Marie Casimire Sobies­ka with her daugh­ter There­sa Kune­gun­da (1690) [160 x 123,5 cm], (26) An uniden­ti­fied painter, Eques­tri­an Por­trait of Marie Casimire Sobies­ka  (approx. 1696) [51 x 40,8 cm], (27) Bartłomiej Stro­bel, Por­trait of Dominik Zasławs­ki-Ostro­gos­ki (1635) [112 x 84 cm], (28) Anony­mous Pol­ish painter, The Sieni­aws­ki fam­i­ly (after 1724) [106 x 144 cm]. Muse­um of King Jan III’s Palace at Wilanów, Warsaw
Fig­ure 164 B. Aes­thet­ic cat­e­go­ry: D. (29) An uniden­ti­fied painter, The por­trait of Prince Ladis­laus Sigis­mund IV Vasa (approx. 1626) [? x ? cm], (30) Anony­mous, Elż­bi­eta Sieni­awska as Min­er­va (before 1725) [141 x 110 cm], (31) Gio­van­ni de Monte, Por­trait of Sebas­t­ian Lubomirs­ki (the sec­ond half of the  16th cen­tu­ry) [176 x 115 cm], (32) Anony­mous, Cof­fin por­trait of Zug­munt Tarło (the end of the 17th cen­tu­ry) [? x ? cm]. Muse­um of King Jan III’s Palace at Wilanów, Warsaw

The divi­sion of images into cat­e­gories is the basis for com­par­ing the exam­ined per­sons behav­iour with regard to time they spent watch­ing them and move­ments of their eye­balls dur­ing both —  the first and sec­ond part of the exper­i­ment. All sta­tis­ti­cal analy­ses of the data col­lect­ed were car­ried out hav­ing in mind the images belong­ing to all four categories.

Global time spent on looking at paintings

After break­ing the images down into four cat­e­gories I ver­i­fied whether the stud­ied peo­ple devot­ed an equal amount of time to watch­ing them. I was also inter­est­ed in how much time the stud­ied peo­ple need­ed to watch the paint­ings in the first part of the exper­i­ment, and how much they need to decide on their aes­thet­ic qual­i­ties in the sec­ond part. For this pur­pose, I con­duct­ed an ano­va vari­ance analy­sis with repeat­ed mea­sure­ment with­in the scope of the vari­able, phase (I and II) and cat­e­go­ry (A, B, C, D) for the glob­al time of watch­ing the paint­ings.  In order to nor­malise the dis­tri­b­u­tion of data, the reac­tion times were sub­ject­ed to log­a­rith­mic transformation. 

First and fore­most it turned that regard­less of the paint­ing cat­e­go­ry, the time devot­ed to watch­ing them was four times longer than in the sec­ond part [F(1, 21) = 195.08; p < 0.001; η2 = 0,90; MPhase I = 9.13 s and MPhase II = 2.39 s].  It also turned out that between the time devot­ed to watch­ing images and their aes­thet­ic pref­er­ence there is a straight-lined cor­re­la­tion: the stud­ied peo­ple spent most time watch­ing those images, which they eval­u­at­ed as the most beau­ti­ful in the sec­ond phase of the exper­i­ment; they devot­ed a lit­tle less time to those which they found to be high­ly aes­thet­ic, and the least — to those which were most rarely eval­u­at­ed as beau­ti­ful [F (3, 63) = 4.90; p < 0.004; η2 = 0.19].

Inter­est­ing­ly, the effect became par­tic­u­lar­ly appar­ent in phase I [F(3, 63) = 10.17; p < 0.001; η2 = 0.33; 165]. Obvi­ous­ly, the dif­fer­ence between the total time of view­ing paint­ings belong­ing to cat­e­gories A and D was sta­tis­ti­cal­ly sig­nif­i­cant [Tukey’s HSD, p < 0.001]. In turn, in phase II, the dif­fer­ences between the cat­e­gories turned out to be insignif­i­cant [F(3, 63) = 2.19; p < 0.098]. It means that the amount of time devot­ed to look­ing at a paint­ing in order to judge it in terms of its aes­thet­ics does not depend on the fact whether the observ­er does or does not like it.

Fig­ure 165. Time of view­ing paint­ings belong­ing to var­i­ous aes­thet­ic categories.

Count and duration total of visual fixation on paintings

Near­ly 86% of the time spent on view­ing the paint­ings exhib­it­ed in the exper­i­ment was spent on visu­al fix­a­tions on its var­i­ous parts. It is there­fore not sur­pris­ing that between the fix­a­tion dura­tion total on paint­ings belong­ing to dif­fer­ent aes­thet­ic cat­e­gories there is an anal­o­gous rela­tion­ship, as between the total view­ing time of these paint­ings and their aes­thet­ic eval­u­a­tion. Sim­i­lar­ly, as in rela­tion to total view­ing time, the analy­sis of ano­va vari­ance with repeat­ed mea­sures of two vari­ables: phase (I, II) and cat­e­go­ry (A, B, C, D) was car­ried out for the fix­a­tion dura­tion total on the whole paintings.

It appeared that the fix­a­tion dura­tion total when view­ing paint­ings in the first phase of the study was dis­pro­por­tion­ate­ly longer than when mak­ing aes­thet­ic deci­sions in phase II [F(1, 18) = 145.86; p < 0.001; η2 = 0.89]. The depen­dence between the fix­a­tion dura­tion total on paint­ings in the first part of the exper­i­ment and the aes­thet­ic pref­er­ence expressed regard­ing them in the sec­ond was con­firmed. The high­er the aes­thet­ic eval­u­a­tion of the paint­ing [F(3, 60) = 10.43; p < 0.001; η2 = 0.34], the longer the dura­tion of fix­a­tion on the paint­ing. Also the dif­fer­ence between the fix­a­tion dura­tion total on paint­ings belong­ing to cat­e­go­ry A and D was sta­tis­ti­cal­ly sig­nif­i­cant [Tukey’s HSD, p < 0.001]. As with total view­ing time for paint­ings in Phase II of the study, there was also no effect of the CATEGORY vari­able for the fix­a­tion dura­tion total [F(3, 57) = 1.76; p = 0.165], and the fix­a­tion dura­tion on paint­ings belong­ing to cat­e­gories A and D was almost iden­ti­cal [Tukey’s HDS, p = 0.99].

The sec­ond gen­er­al para­me­ter char­ac­ter­is­ing eye­ball move­ment, apart from fix­a­tion dura­tion total, was the count of fix­a­tion points on the analysed paint­ing. Just like the fix­a­tion dura­tion total, the count of fix­a­tion points is high­ly cor­re­lat­ed with the total image view­ing time. For this rea­son, the count of fix­a­tion points dur­ing phase I (watch­ing) was sig­nif­i­cant­ly high­er than the count of fix­a­tion points dur­ing phase II [F(1, 21) = 50.91; p < 0.001; η2 = 0.74]. While view­ing paint­ings in the first part of the exper­i­ment, the stud­ied peo­ple more often fixed their vision on the paint­ings belong­ing to the high­er aes­thet­ic cat­e­gories [F(3, 60) = 10.55; p < 0.001; η2 = 0.35]. The dif­fer­ence between the count of fix­a­tion points while casu­al­ly view­ing paint­ings belong­ing to cat­e­go­ry A and D was sta­tis­ti­cal­ly sig­nif­i­cant [Tukey’s HSD, p < 0.001].

In the sec­ond part of the exper­i­ment, in which the respon­dents made an aes­thet­ic eval­u­a­tion, an inter­est­ing effect emerged, which I did not find in the analy­sis of glob­al view­ing time. At this stage of the study, there were sig­nif­i­cant dif­fer­ences in the count of fix­a­tion points on paint­ings belong­ing to dif­fer­ent aes­thet­ic cat­e­gories [F(3, 63) = 3.25; p = 0.027; η2 = 0.13; Fig.  166]. How­ev­er, this effect was not affect­ed by the dif­fer­ence between A and D cat­e­gories, which I have so far observed for view­ing times in Phase I, but by the dif­fer­ence between B and D cat­e­gories (Tukey’s HSD, p = 0.018). The high­est count of fix­a­tion points appeared in paint­ings that belonged to cat­e­go­ry B, i.e. not the most beau­ti­ful ones, but the inclu­sion of which in cat­e­go­ry A could be par­tic­u­lar­ly dif­fi­cult for the respon­dents. The increase in the count of fix­a­tion points on paint­ings includ­ed in this cat­e­go­ry was most prob­a­bly relat­ed to the uncer­tain­ty of the respon­dents as to their aes­thet­ic val­ue. They were con­front­ed with the need to make a deci­sion and pre­ferred to once again “take a look” at more ele­ments in these paint­ings in order to make the right deci­sion. An anal­o­gous, although sta­tis­ti­cal­ly insignif­i­cant, result appeared also in rela­tion to the fix­a­tion dura­tion total.

Fig­ure 166. Count of fix­a­tion points on paint­ings belong­ing to dif­fer­ent aes­thet­ic cat­e­gories dur­ing their aes­thet­ic eval­u­a­tion (phase II).

The para­me­ter con­cern­ing the count of fix­a­tion points may also be expressed through fix­a­tion fre­quen­cy, which means the count of fix­a­tion points over one sec­ond. As it turns out, with­in all cat­e­gories, fix­a­tion fre­quen­cy is sig­nif­i­cant­ly high­er dur­ing aes­thet­ic eval­u­a­tion in the sec­ond phase of the exper­i­ment than while view­ing the paint­ings in phase I [F(1, 18) = 15.63; p < 0.001; η2 = 0.47]. In oth­er words, ocu­lo­mo­tor activ­i­ty increas­es sig­nif­i­cant­ly dur­ing aes­thet­ic eval­u­a­tion. How­ev­er, no dif­fer­ences were iden­ti­fied in the fix­a­tion fre­quen­cy in rela­tion to dif­fer­ent aes­thet­ic cat­e­gories, both in Phase I and Phase II. This result con­firms the pre­sent­ed inter­pre­ta­tion con­cern­ing the increased fix­a­tion count in Cat­e­go­ry B in Phase II. Deci­sions con­cern­ing the aes­thet­ic val­ue of a paint­ing are accom­pa­nied by inten­si­fied ocu­lo­mo­tor activ­i­ty result­ing from slight­ly increased emo­tion­al stimulation.

Anoth­er indi­ca­tor relat­ed to visu­al fix­a­tions on a paint­ing is fix­a­tion dura­tion aver­age. It con­sti­tutes the quo­tient of fix­a­tion dura­tion total divid­ed by their count. This indi­ca­tor is par­tic­u­lar­ly inter­est­ing due to the eval­u­a­tion of sen­so­ry pro­cess­ing strength lev­el.  How­ev­er, it turned out that the fix­a­tion dura­tion aver­age was more or less the same dur­ing the first and the sec­ond phase of the exper­i­ment [F(1, 21)= 3.48; p=0.076], as well as in rela­tion to all aes­thet­ic cat­e­gories, both dur­ing casu­al view­ing of the paint­ings [F(3, 63) = 0.56; p = 0.644] and dur­ing the aes­thet­ic eval­u­a­tion phase [F(3, 63) = 0.21; p = 0.887]. Of course, the dif­fer­ences between extreme aes­thet­ic cat­e­gories in both phas­es of the exper­i­ment were also sta­tis­ti­cal­ly insignif­i­cant [Tukey’s HSD, phase I – p=0.999 and phase II – 0,930]. In oth­er words, the lev­el of engage­ment in the pro­cess­ing of data cap­tured by the reti­nae of sub­jects was more or less the same regard­less of the phase of the exper­i­ment and the cat­e­go­ry of the viewed paintings.

How did the sight travel across beautiful portraits, and how did it travel across the non-beautiful ones?

The sec­ond group of para­me­ters of the eye­ball move­ment tra­jec­to­ry are the sac­cades. Their objec­tive is to shift the visu­al axis from one part of the paint­ing to anoth­er. Sac­cades are mea­sured in units of length, e.g. pix­els locat­ed between two fix­a­tion points or mil­lime­tres, or in angu­lar units. The dis­tance between two fix­a­tion points is called the sac­cade ampli­tude. The sum of all sac­cade ampli­tudes record­ed dur­ing view­ing the entire paint­ing deter­mines the length of the scan­path. With this para­me­ter I will begin to dis­cuss the tra­jec­to­ry of eye­ball move­ment while view­ing Wilanów portraits.

Because of the much greater amount of time devot­ed to view­ing images in the first part of the exper­i­ment, it is under­stand­able that in this phase the eyes of the respon­dents “fol­lowed” much longer than in phase II. This obvi­ous intu­ition is con­firmed by the results of vari­ance analy­sis [F(1, 21) = 192.00; p < 0.001; η2 = 0.90], in which, as before, phase (I and II) and cat­e­go­ry (A, B, C, D) con­sti­tut­ed the inde­pen­dent vari­ables, but this time the length of the scan­ning path expressed in log­a­rithms of the num­ber of screen pix­els that divid­ed the sub­se­quent fix­a­tion points were the depen­dent vari­able. The use of log­a­rith­mic trans­for­ma­tion in rela­tion to var­i­ous para­me­ters of eye­ball move­ments not only nor­malis­es their dis­tri­b­u­tion, but also sig­nif­i­cant­ly reduces indi­vid­ual dif­fer­ences between the respon­dents, thus increas­ing the inter­nal homo­gene­ity of the com­pared groups.

It also turned out that the total length of the scan­path was depen­dent on which cat­e­go­ry the paint­ing belonged to [F(3, 63) = 7.35; p < 0.001; η2 = 0.26; Fig. 167]. Dif­fer­ences between cat­e­gories A and C as well as B and C were sta­tis­ti­cal­ly sig­nif­i­cant [Turkey’s HSD, p < 0.006 and p < 0.003 respec­tive­ly]. This effect was sim­i­lar­ly vis­i­ble in both phas­es of the experiment.

Fig­ure 167. Scan­path length of paint­ings belong­ing to dif­fer­ent aes­thet­ic categories

This result is worth not­ing for two rea­sons. First of all, it shows that more beau­ti­ful paint­ings (A and B) are viewed with greater involve­ment than paint­ings belong­ing to the oth­er two cat­e­gories. To a large extent, this result is a deriv­a­tive of the longer dura­tion and greater count of visu­al fix­a­tions on these paint­ings. Sec­ond­ly, the biggest dif­fer­ence in terms of scan­path length is not in the case of those paint­ings that fall under cat­e­go­ry A or D, but those that are in between, i.e. B and C. It seems that cat­e­go­ry B paint­ings required the most atten­tion from the respon­dents, while cat­e­go­ry C paint­ings required the least of all. Aes­thet­ic eval­u­a­tion of the kind: “rather beau­ti­ful” seems more dif­fi­cult than “rather non- beautiful”.

Anoth­er mea­sure of sac­cades based on their length is their aver­age ampli­tude. This is the quo­tient of the scan path length by the num­ber of sac­cades it con­tains. There were no sta­tis­ti­cal­ly sig­nif­i­cant dif­fer­ences between the first and sec­ond phase of the exper­i­ment [F(1, 21) = 0.5; p = 0.487]. Aver­age ampli­tude of sac­cades was sim­i­lar in both phas­es of the study.

The vari­able that influ­enced the aver­age ampli­tude of the sac­cades was cat­e­go­ry [F(3, 63) = 9.17; p < 0.001; η2 = 0.30; Fig. 168]. The effect in the same form was observed in both phas­es of the exper­i­ment. In con­trast to the scan­path length, it was found that the aver­age ampli­tudes of the sac­cades when view­ing cat­e­go­ry A paint­ings, i.e. the most beau­ti­ful ones, were sig­nif­i­cant­ly short­er than the aver­age ampli­tudes of the sac­cades when view­ing cat­e­go­ry D paint­ings [Tuykey’s HSD, p = 0.017].

Fig­ure 168. Aver­age length of sac­cades when view­ing images belong­ing to dif­fer­ent aes­thet­ic categories

Such a result would sug­gest that in the case of more beau­ti­ful paint­ings, there is a ten­den­cy for more con­cen­trat­ed view­ing of frag­ments of the paint­ings (the fix­a­tion points are locat­ed clos­er to each oth­er), while in the case of “reject­ed” paint­ings, the ten­den­cy for more scat­tered con­cen­tra­tion of vision on dis­tant frag­ments is more pro­nounced. With­out elab­o­rat­ing on this issue fur­ther here, I would just like to remind that such a view­ing strat­e­gy is more char­ac­ter­is­tic for experts than novices. I will be look­ing at this issue again in the dis­cus­sion of the results.

Each sac­cade is per­formed at a spe­cif­ic time. The aver­age dura­tion of the sac­cade is anoth­er indi­ca­tor which I took into account in the analy­sis of ocu­lo­mo­tor data. This is a very sub­tle para­me­ter of the eye move­ment tra­jec­to­ry, con­sid­er­ing that its length is about 1/10 of a sec­ond. It turned out, how­ev­er, that twice as much time of vision shift was record­ed when view­ing paint­ings in phase I as dur­ing the aes­thet­ic eval­u­a­tion in phase II [F(1, 18) = 121,01; p < 0.001; η2 = 0.87; MFaza I = 71.3 ms, MFaza II = 159.5 ms] and for paint­ings more beau­ti­ful (cat­e­gories A and B) than less beau­ti­ful (C and D) [F(3, 54) = 6.34; p < 0.001; η2 = 0.26; Fig. 169]. The dif­fer­ences between cat­e­gories B and C and B and D are also sta­tis­ti­cal­ly sig­nif­i­cant [Tukey’s HSD, p < 0.001 and p = 0.019 respectively].

Fig­ure 169. Aver­age time of sac­cades when view­ing images belong­ing to dif­fer­ent aes­thet­ic categories

The last of the analysed mea­sures of sac­cades was their speed. As before, in rela­tion to this para­me­ter I also used the analy­sis of vari­ance with repeat­ed phase (I and II) and cat­e­go­ry (A, B, C and D) vari­ables. It turned out that the sac­cades were much faster when view­ing images (MPha­seI = 396o/s) than dur­ing their aes­thet­ic eval­u­a­tion (MPhase II = 290o/s) [F(1, 21) = 87.47; p < 0.001; η2 = 0.81]. There were also dif­fer­ences in the speed of the sac­cades when view­ing paint­ings belong­ing to dif­fer­ent aes­thet­ic cat­e­gories [F(3, 63) = 11.98; p < 0.001; η2 = 0.36]. The depen­den­cies between the cat­e­gories were sim­i­lar to those of the scan­path length, which indi­cates a high cor­re­la­tion between the two para­me­ters (Fig. 167).

Areas of major interest

The analy­sis of the eye­ball move­ment tra­jec­to­ry can be per­formed either for the whole paint­ing or for a part of it which for some rea­son is of inter­est to the researcher or aroused spe­cial inter­est of the respon­dents. In all the pre­sent­ed paint­ings, the dom­i­nant ele­ment was a man. No won­der that no mat­ter which aes­thet­ic cat­e­go­ry a giv­en paint­ing belonged to, the faces of peo­ple present in it aroused the great­est curios­i­ty. They were the main focus of the respon­dents’ glances, which is best reflect­ed in the atten­tion maps (Fig.  170 A and B).

Fig­ure 170 A. Atten­tion map plot­ted on the paint­ing from Wilanow col­lec­tion: An uniden­ti­fied Pol­ish painter, The Sieni­aws­ki Fam­i­ly (after 1724) [106 x 144 cm]
Fig­ure 170 B. Atten­tion map plot­ted on the paint­ing from Wilanow col­lec­tion: Domeni­co Zampieri called Domenichi­no, Guardian Angel [132 x 100 cm]

It should be remind­ed that the inten­si­ty of atten­tion mea­sured by the count and dura­tion of fix­a­tion is pre­sent­ed graph­i­cal­ly in Fig. 170 A and B by means of col­ors, where red means the great­est lev­el of inter­est of a giv­en frag­ment of the paint­ing, then yel­low, green, light and dark blue to fuch­sia, i.e. a mix­ture of vio­let and pink. The remain­ing, dark areas mean those places that did not arouse the exam­ined per­sons’ interest.

Based on atten­tion maps, an ellip­ti­cal shape with the same chords was estab­lished for all the paint­ings, cov­er­ing the area or areas there­of that were of par­tic­u­lar inter­est to the exam­ined per­sons. It was the so called area of inter­est (AOI). This area sur­face in rela­tion to sur­face of the screen — thanks to which the image was dis­played was approx.  2%. It was so large, that it cov­ered every face inde­pen­dent­ly from its size (com­pare Fig. 171 A, B and C). The cri­te­ri­on for deter­min­ing length of AOI chords was cov­er­age of the atten­tion map with­in the range from red to dark blue, inclu­sive. The exam­ples of images showed in Fig. 170 and Fig. 171 leave no doubt that if they include faces, then the most time is devot­ed to them in com­par­i­son to all the oth­er things in the picture.

Fig­ure 171. Three sam­ple paint­ings depict­ing char­ac­ters with dif­fer­ent face sizes: A — large, B — medi­um, C — small, with super­im­posed areas of the high­est inter­est, with the same dimen­sions (the yel­low ellipse).

Glances at the faces of the portrayed people and beyond them

The ocu­lo­mo­tor data on the areas of the high­est inter­est (AOI) as well as the rest of the images (non-AOI) can be analysed anal­o­gous­ly as with regard to the entire images.  The break­down of paint­ing space into AOI and non-AOI also makes it pos­si­ble to con­duct addi­tion­al com­par­isons between the images belong­ing to var­i­ous aes­thet­ic cat­e­gories.  You can, for exam­ple, find out what role in the aes­thet­ic eval­u­a­tion was played by the aes­thet­ic eval­u­a­tion of the face alone.  Are some paint­ings found to be beau­ti­ful because the faces of the depict­ed char­ac­ters are good-look­ing, and oth­ers are not regard­ed as beau­ti­ful because the faces in them are ugly?  If there was a sig­nif­i­cant cor­re­la­tion between the aes­thet­ic eval­u­a­tion of an entire image and the face depict­ed in it, we could expect that the vari­able cat­e­go­ry would sim­i­lar­ly influ­ence the ocu­lo­mo­tor para­me­ters for AOI, name­ly the faces, as for the entire paint­ings.  In order to answer these ques­tions I con­duct­ed ano­va vari­ance analy­ses, sim­i­lar to those con­duct­ed before — for vari­ables: phase (I and II) and cat­e­go­ry (A, B, C, D), sep­a­rate­ly for AOI and non-AOI. 

The first para­me­ter, which is worth con­sid­er­ing in the analy­sis of the tra­jec­to­ry of eye­ball move­ment in a dis­tin­guished area of the paint­ing is the fix­a­tion dura­tion total, which makes it pos­si­ble to get the gen­er­al idea regard­ing the inter­est in the area. As a result of the con­duct­ed analy­sis it turned out that the stud­ied per­sons fixed their vision on faces, name­ly AOI, in phase I for much longer than while per­form­ing the aes­thet­ic eval­u­a­tion of the paint­ings in phase II  [F(1, 21) = 127,81; p < 0,001; η2 = 0,86]. As with the AOI, also out­side the AOI, sight was fixed on images in phase I much longer than in phase II [F(1, 21) = 139.78; p < 0.001; η2 = 0.87]. These results only reflect the more gen­er­al ten­den­cy, which was already found in the analy­ses of whole paint­ings, accord­ing to which the fix­a­tion dura­tion total was longer in the first part of the study than in the sec­ond.  In this sense, the result does not add any­thing new to what we already know. 

The results of the analy­ses of fix­a­tion dura­tion total with regard to the aes­thet­ic cat­e­gories are far more inter­est­ing. If we take a good look at the amount of time devot­ed by the par­tic­i­pants to the faces on the por­traits, name­ly AOI, we will come to real­ize that, indeed, there are sta­tis­ti­cal­ly sig­nif­i­cant dif­fer­ences between them [F(3, 63) = 6.95; p < 0.001; η2 = 0.25; fig. 172].

Fig­ure 172. Fix­a­tion dura­tion total on the faces of peo­ple whose por­traits were clas­si­fied into var­i­ous aes­thet­ic categories.

The data pre­sent­ed in Fig. 172, how­ev­er, do not reflect the depen­dence: the more beau­ti­ful is the paint­ing, the longer it is viewed.  On the con­trary, the longest views, mea­sured by the fix­a­tion dura­tion total on the faces of the mod­els, were record­ed in the case of the paint­ings belong­ing to cat­e­go­ry C, that is those, the aes­thet­ic qual­i­ties of which raised rather more than less doubts. But, first and fore­most, it turned out that the dif­fer­ences in terms of time devot­ed to look­ing at faces in pic­tures belong­ing to cat­e­gories A and D were sta­tis­ti­cal­ly insignif­i­cant [Tukey’s HSD, p= 0.756].  The same effect was observed in both phas­es of the exper­i­ment, which only rein­forces the hypoth­e­sis that the aes­thet­ic eval­u­a­tion of the paint­ings is not affect­ed by the aes­thet­ic eval­u­a­tion of the faces of the por­trayed people.

The analy­sis of the area out­side the AOI revealed an even more inter­est­ing effect. When view­ing paint­ings eval­u­at­ed as beau­ti­ful, the respon­dents devot­ed much more time to areas out­side the AOI than when view­ing paint­ings belong­ing to the cat­e­go­ry of non-beau­ti­ful [F(3, 63) = 9.99; p < 0.001; η2 = 0.30; Fig. 173]. The dif­fer­ences between the paint­ings belong­ing to the more pos­i­tive­ly eval­u­at­ed cat­e­gories, i.e. A and B, and the paint­ings belong­ing to the more neg­a­tive­ly eval­u­at­ed cat­e­gories, i.e. C and D, are sta­tis­ti­cal­ly sig­nif­i­cant [Tukey’s HSD, A:C — p < 0.001, A:D — p = 0.015, B:C — p < 0.001, B:D — p = 0.017].

Fig­ure 173. Fix­a­tion dura­tion total on areas of paint­ings clas­si­fied into var­i­ous aes­thet­ic cat­e­gories, which lie out­side the AOI, i.e. out­side the faces of the por­trayed persons.

It is also note­wor­thy that the depen­den­cies pre­sent­ed in Fig. 173 refer main­ly to the first part of the exper­i­ment, dur­ing which the respon­dents with­out any time lim­its viewed the paint­ings pre­sent­ed to them [F(3, 63) = 18.51; p < 0.001; η2 = 0.47]. On the oth­er hand, dur­ing mak­ing deci­sions regard­ing the inclu­sion of a giv­en paint­ing into cat­e­gories of beau­ti­ful or non-beau­ti­ful, the total amount of time devot­ed to view­ing areas that lie beyond the AOI was the same, inde­pen­dent­ly to which cat­e­go­ry a giv­en paint­ing was final­ly clas­si­fied [F(3, 63) = 1.99; p = 0.123].

In order to under­stand the spe­cif­ic nature of a much longer time of view­ing non-AOI areas in beau­ti­ful paint­ings com­pared to the same areas in non-beau­ti­ful paint­ings, it is worth to exam­ine two pairs of paint­ings belong­ing to these cat­e­gories. Their sim­i­lar­i­ty in com­po­si­tion was assumed as the basis for their selec­tion to illus­trate the effect. The first pair are horse por­traits of Stanisław Kost­ka Potoc­ki and Maria Kaz­imiera Sobies­ka (Fig. 174), the sec­ond one — of the bust of Stanisław Kost­ka Potoc­ki and Zyg­munt Tarło (Fig. 175). The atten­tion map was plot­ted on each of the paint­ings. It reveals the loca­tion and time of the respon­dents’ looks on dif­fer­ent parts of these paintings.

Fig­ure 174. Paint­ings belong­ing to two aes­thet­ic cat­e­gories (A and D) with atten­tion map plot­ted on the paint­ings, A — Jacques Loius David, The Horse Por­trait of Count Stanis­las Potoc­ki (1781) and D — Anony­mous Painter, The Horse Por­trait of Maria Kaz­imiera Sobies­ka (around 1696). First example
Fig­ure 175. Paint­ings belong­ing to two aes­thet­ic cat­e­gories (A and D) with atten­tion map plot­ted on the paint­ings, A — Anton Graff, The Por­trait of Count Stanis­las Potoc­ki (1785) and D — Anony­mous, Cof­fin por­trait of Zug­munt Tarło (the end of the 17th cen­tu­ry). Sec­ond example

Both exam­ples clear­ly show that, as I have point­ed out on mul­ti­ple occa­sions, the ele­ment that attracts the most of the view­er’s atten­tion is the face of the por­trayed per­son (as indi­cat­ed by the red dots). How­ev­er, it is not the dura­tion of view­ing the faces that deter­mines the sig­nif­i­cant­ly longer total time devot­ed to view­ing beau­ti­ful paint­ings as opposed to the non-beau­ti­ful ones. This is deter­mined by the dura­tions of look­ing at oth­er parts of the paint­ings, as indi­cat­ed by the bright-blue spots marked by arrows of the same color.

It is clear at first glance that the loca­tions marked by brighter spots are present only in the paint­ings cat­e­gorised as beau­ti­ful. They are the sources of dif­fer­ences in the total time of view­ing beau­ti­ful and non-beau­ti­ful pieces of art. In Fig. 176, there are four parts of beau­ti­ful paint­ings which attract­ed view­ers’ spe­cial atten­tion. What do these parts con­tain? In two cas­es (A and B), they are parts of a rich­ly-embell­ished horse har­ness, and in one case, it is a matt-shine part of a medal on Count Stanis­las Potock­i’s chest (D). A com­mon fea­ture of these parts is their shim­mer and con­trastive nature, which, how­ev­er, are not enough to unequiv­o­cal­ly repro­duce the shapes of the paint­ed objects. In this respect, they are mys­te­ri­ous; they attract the view­er’s atten­tion and, at the same time, remain unseen in their full glory.

Fig­ure 176. Details out­side the AOI that attract­ed par­tic­u­lar­ly high atten­tion from the view­ers: A, B and C – parts of the eques­tri­an por­trait of Count Stanis­las Potoc­ki, D – part of Por­trait of Count Stanis­las Potocki

The trait of mys­te­ri­ous­ness stem­ming from ambi­gu­i­ty may also be ascribed to the fourth of the select­ed parts ©. It depicts the head of a hunt­ing dog, seem­ing­ly in attack posi­tion, but at the same time filled with fear in the pres­ence of a huge horse whose hoof is much too close to the dog’s muz­zle  What is about to hap­pen here? It appears that the horse is more inter­est­ed in the dog than in the rid­er on its back. If the horse kicks the dog, it will undoubt­ed­ly retal­i­ate with even more fury. What will then hap­pen to the rid­er? This paint­ing con­sti­tutes the last frame before the one in which we can expect sub­stan­tial change in the action.

Roland Barthes (2008) refers to such parts (details) in a paint­ing, which build ten­sion through their ambi­gu­i­ty or uncer­tain­ty, as the punc­tum, as opposed to the studi­um, i.e. the fun­da­men­tal sub­ject of a paint­ing. The sub­ject of Jacques-Louis David’s paint­ing is the por­trait of Count Stanis­las Potoc­ki on a horse, who salutes the view­er with a stud­ied ges­ture. How­ev­er, Potoc­ki is hard­ly the ele­ment which builds ten­sion in this scene. It is that very part in the bot­tom left cor­ner which attracts the view­ers’ attention. 

A few more words about visual fixations on AOI and non-AOI

After this brief qual­i­ta­tive analy­sis of the respon­dents’ areas of spe­cial inter­est in beau­ti­ful paint­ings, now is the time to refer to oth­er results of sta­tis­ti­cal analy­sis of ocu­lo­mo­tor data. To a large extent, they con­firm the results of analy­ses of the fix­a­tion dura­tion total on the paintings.

As with the fix­a­tion dura­tion total on the AOI and on the non-AOI, also the count of fix­a­tion points on both these areas was greater dur­ing the first phase of the exper­i­ment than dur­ing the sec­ond phase [AOI: F(1, 21) = 43.45; p < 0.001; η2 = 0.67 and non-AOI: F(1, 21) = 50.70; p < 0.001; η2 = 0.71]. This result does not require any addi­tion­al comment.

Also, the count of fix­a­tion points in the AOI turned out to be slight­ly dif­fer­ent depend­ing on the aes­thet­ic cat­e­go­ry to which the paint­ing belonged [F(3, 63) = 9.81; p < 0.001; η2 = 0.32]. How­ev­er, the speci­fici­ty of this dif­fer­en­ti­a­tion was unique. It appeared that the count of fix­a­tion points in the AOI was very sim­i­lar for the three cat­e­gories of paint­ings: A, B and C, i.e. both beau­ti­ful, rather beau­ti­ful and rather non-beau­ti­ful. In con­trast, the count of fix­a­tion points on the faces of por­traits con­sid­ered to be the least beau­ti­ful was sig­nif­i­cant­ly low­er. There­fore, we can­not refer to any unam­bigu­ous influ­ence of a cat­e­go­ry vari­able on the count of fix­a­tion points on beau­ti­ful and non-beau­ti­ful faces. A sim­i­lar phe­nom­e­non also appeared in rela­tion to the fix­a­tion dura­tion total on the faces of the por­trayed persons.

In turn, apart from the AOI, the results are more unam­bigu­ous and con­firm the ear­li­er find­ings. The count of fix­a­tion points for non-AOI was found to dif­fer depend­ing on the aes­thet­ic cat­e­go­ry [F(3, 63) = 19.06; p < 0.001; η2 = 0.48]. In the paint­ings belong­ing to cat­e­gories A and B, the respon­dents placed much more fix­a­tion points than in the paint­ings belong­ing to cat­e­gories C and D. As with fix­a­tion dura­tion total, all these dif­fer­ences in pairs are sta­tis­ti­cal­ly sig­nif­i­cant [Tukey’s HSD, A:C — p < 0.001, A:D — p < 0.001, B:C — p < 0.001, B:D — p = 0.039].

And final­ly, one more para­me­ter of the eye­ball move­ment tra­jec­to­ry, i.e. the fix­a­tion dura­tion aver­age on AOI and non-AOI. In rela­tion to the faces of the por­trayed per­sons, i.e. AOI, there were no sta­tis­ti­cal dif­fer­ences between the phas­es of study [F(1, 21) = 5.15; p = 0.090] and between the aes­thet­ic cat­e­gories [F(3, 63) = 1.73; p = 0.169] due to the fix­a­tion dura­tion aver­age. In short, the lev­el of pro­cess­ing of sen­so­ry data col­lect­ed dur­ing visu­al fix­a­tion on the faces of all por­trayed per­sons was sim­i­lar. This result leaves no doubt that the “aes­thet­ic qual­i­ty” of the face had no influ­ence on the aes­thet­ic eval­u­a­tion of the entire paintings.

As for non-AOI, the data on fix­a­tion dura­tion aver­age sig­nif­i­cant­ly sup­ports the hypoth­e­sis of the impor­tance of these areas in the aes­thet­ic eval­u­a­tion of paint­ings. A much deep­er lev­el of sen­so­ry data pro­cess­ing accom­pa­nied the view­ing of all paint­ings in the first part of the study than in the sec­ond part [F(1, 21) = 58.06; p < 0.001; η2 = 0.73]. It was also dif­fer­ent depend­ing on the aes­thet­ic cat­e­go­ry to which the giv­en paint­ing belonged [F(3, 63) = 5.34; p = 0.002; η2 = 0.20; Fig. 177]. Of course, the dif­fer­ence between cat­e­go­ry A and D is sta­tis­ti­cal­ly sig­nif­i­cant [Tukey’s HSD, p = 0.002].

Fig­ure 177. The fix­a­tion dura­tion aver­age on the areas of paint­ings clas­si­fied into dif­fer­ent aes­thet­ic cat­e­gories, which lie out­side the AOI, i.e. out­side the faces of the por­trayed persons

What happens on the border between the AOI and beyond?

While view­ing a paint­ing, the observ­er directs his or her gaze to dif­fer­ent parts of it and often returns to the places he or she has already seen. They are the so-called regres­sive sac­cades. If a par­tic­u­lar frag­ment, e.g. the AOI, is iden­ti­fied in the paint­ing, then the regres­sive sac­cades are an indi­ca­tor of a kind of “bor­der move­ment” between this area and the areas out­side it. The num­ber of regres­sive sac­cades reg­is­tered at the bor­der of the AOI may indi­cate a need to explore this area again as well as a need to explore an area out­side it. The inter­pre­ta­tion to be adopt­ed depends on the inter­pre­ta­tion of oth­er para­me­ters of the eye­ball move­ment trajectory.

For obvi­ous rea­sons, the num­ber of regres­sive sac­cades in the first phase of the exper­i­ment was much greater than in the sec­ond phase [F(1, 21) = 19.42; p < 0.001; η2 = 0.48]. First of all, it results from the fact that the ocu­lo­mo­tor activ­i­ty dur­ing free view­ing of paint­ings in phase I is much high­er than dur­ing their aes­thet­ic eval­u­a­tion in phase II. Also inter­est­ing are the dif­fer­ences in the num­ber of regres­sive sac­cades in rela­tion to paint­ings belong­ing to dif­fer­ent aes­thet­ic cat­e­gories. It turns out that the num­ber of regres­sive sac­cades depends on which aes­thet­ic cat­e­go­ry a giv­en paint­ing belongs to [F(3, 63) = 4,58; p = 0,006; η2 = 0,18]. When view­ing the most beau­ti­ful cat­e­go­ry A paint­ings, there were many more regres­sive sac­cades than when view­ing cat­e­go­ry D paint­ings [Tukey’s HSD, p = 0.020]. This proves the high ocu­lo­mo­tor dynam­ics when view­ing beau­ti­ful paint­ings, com­pared to non-beau­ti­ful ones.

Observing paintings in time

In the analy­ses pre­sent­ed, I have tak­en into account dif­fer­ent aspects of time spent on view­ing the paint­ings: total time deter­mined by the image expo­sure time, as well as fix­a­tion / sac­cade dura­tion total / aver­age. None of these ocu­lo­mo­tor para­me­ters has been analysed so far, tak­ing into account dynam­ics of changes in the sub­se­quent stages of look­ing at paint­ings over time. Sup­pose one is look­ing at a paint­ing for 30 sec­onds. Is the fix­a­tion count, fix­a­tion dura­tion aver­age or sac­cade length the same at the begin­ning, in the mid­dle and at the end of this time? Is the pos­si­ble dynam­ics of changes with­in the scope of these para­me­ters the same, with regard to beau­ti­ful and as non-beau­ti­ful paintings?

In order to answer these ques­tions, the glob­al time spent observ­ing every paint­ing by each exam­ined per­son was divid­ed into four quar­ters, here­inafter referred to as “paint­ing obser­va­tion stages”. Of course, the length of indi­vid­ual stages dif­fered slight­ly depend­ing on exam­ined per­son and paint­ing, but on aver­age it fluc­tu­at­ed around approx. 2 s [SD = 1.54]. Then, the fol­low­ing val­ues of three ocu­lo­mo­tor para­me­ters were esti­mat­ed for each phase: the fix­a­tion dura­tion aver­age, the sac­cade length aver­age and the sac­cade dura­tion aver­age that under­went the analy­sis of ano­va vari­ance with repeat­ed mea­sures of two inde­pen­dent vari­ables: cat­e­go­ry (A, D) and phase (1, 2, 3 and 4) In these analy­ses I lim­it­ed myself to only two extreme aes­thet­ic cat­e­gories and to the first phase of the exper­i­ment. Above all, I was inter­est­ed in whether, based on the ocu­lo­mo­tor data from the sub­se­quent stages of view­ing the paint­ings in the first part of the exper­i­ment, an aes­thet­ic deci­sion could be pre­dict­ed, in the sec­ond part.

The analy­sis of the fix­a­tion dura­tion aver­age has deliv­ered some puz­zling results. It turned out that between suc­ces­sive stages of observ­ing the paint­ings, sta­tis­ti­cal­ly sig­nif­i­cant dif­fer­ences exist [F(3, 63) = 12.88; p < 0.001; η2 = 0.38]. On the oth­er hand, at the lev­el of the prin­ci­pal effects, dif­fer­ences between cat­e­gories were insignif­i­cant, what con­firms the results of ear­li­er analy­ses of this para­me­ter. How­ev­er, the most inter­est­ing was the result of inter­ac­tive influ­ence of both inde­pen­dent vari­ables on the fix­a­tion dura­tion aver­age [F(3, 63) = 2.68; p = 0.054; η2 = 0.11; Fig. 178).

Fig­ure 178. Inter­ac­tion between vari­ables AESTHETIC CATEGORY and PAINTING VIEWING STAGE in rela­tion to the fix­a­tion dura­tion aver­age in the first phase of the experiment.

The longer fix­a­tion dura­tion aver­age in the first phase of view­ing, was short­ened in phase 2 and 3 and then returned to the ini­tial val­ue in phase 4. Nev­er­the­less, a dif­fer­ence between fix­a­tion dura­tion aver­ages on paint­ings belong­ing to both extreme aes­thet­ic cat­e­gories at phase 1 [Tukey’s HSD, p = 0.038]. It indi­cates that already dur­ing the first 2 sec­onds after stim­u­lus expo­sure, the exam­ined per­sons devot­ed much less time to analy­sis of sen­so­ry data from sub­se­quent fix­a­tion points on these paint­ings which were indi­cat­ed as beau­ti­ful in the sec­ond phase of the exper­i­ment, com­pared with fix­a­tion dura­tion aver­ages on paint­ings that were regard­ed as non-beau­ti­ful in the sec­ond phase of the experiment.

How is that pos­si­ble? Could it be that the moment of aes­thet­ic eval­u­a­tion of a work of art being viewed is ahead its more accu­rate, visu­al analy­sis?  Is it pos­si­ble that the visu­al data col­lect­ed dur­ing the sub­se­quent phas­es of view­ing the paint­ing do not influ­ence its lat­er eval­u­a­tion? The pre­sent­ed results of the analy­ses of fix­a­tion dura­tion aver­age shed an inter­est­ing light on the moment of the emer­gence of aes­thet­ic judge­ment. The assump­tion, accord­ing to which the aes­thet­ic judge­ment is pre­ced­ed by a deep­er analy­sis of a work of art, under­laid the pre­vi­ous analy­ses.  It is pos­si­ble, how­ev­er, that it is pre­cise­ly the oth­er way around. 

In Fig. 178 I also marked graphs of two lin­ear func­tions, best adjust­ed to the fix­a­tion dura­tion aver­ages record­ed for sub­se­quent phas­es of view­ing paint­ings belong­ing to cat­e­gories A and D. With regard to the paint­ings belong­ing to cat­e­go­ry A, name­ly eval­u­at­ed as beau­ti­ful, an upward trend (thin blue line) can be found, where­as in the case of paint­ings belong­ing to cat­e­go­ry D, the fix­a­tion dura­tion aver­age becomes short­er dur­ing the course of paint­ing view­ing (thin red line).

Using the vari­ance analy­sis I also analysed the influ­ence of the inde­pen­dent vari­ables: cat­e­go­ry (A and D) and phase (1, 2, 3 and 4) on the aver­age length of sac­cades and the aver­age time of their dura­tion.  It turned out that the aver­age sac­cade length increased dur­ing the view­ing [F(3, 63) = 21.89; p < 0.001; η2 = 0.51; Fig. 179], just like the aver­age time of their dura­tion [F3, 63 = 0.71; p = < 0.001; η2 = 0.32; Fig.  180]. Obvi­ous­ly, both depen­dent vari­ables have turned out to be vul­ner­a­ble to the impact of the vari­able ‘cat­e­go­ry,’ which was the sub­ject of analy­sis ear­li­er (cf. Fig. 168 and Fig. 169). How­ev­er, there was no inter­ac­tion effect between the vari­ables ‘cat­e­go­ry’ and ‘view­ing stage’ in rela­tion to these depen­dent variables.

Fig­ure 179. Inter­ac­tion between vari­ables AESTHETIC CATEGORY and IMAGE VIEWING STAGE in rela­tion to the aver­age sac­cade length in the first phase of the experiment
Fig­ure 180. Inter­ac­tion between vari­ables AESTHETIC CATEGORY and IMAGE VIEWING STAGE in rela­tion to the aver­age sac­cade time in the first phase of the experiment

Short summary and discussion of results

The pre­sent­ed analy­ses of ocu­lo­mo­tor data of peo­ple who viewed and then eval­u­at­ed paint­ings in terms of aes­thet­ics reveal many inter­est­ing reg­u­lar­i­ties, allow­ing for opti­mistic out­look regard­ing the pos­si­bil­i­ty of using ocu­log­ra­phy tech­nique to study even such sub­tle cat­e­gories as beau­ty. On the one hand, it turns out that on the basis of para­me­ters con­cern­ing the eye move­ment tra­jec­to­ry, one can pre­dict the aes­thet­ic pref­er­ences of the sub­jects. Fur­ther­more, these data con­sti­tute empir­i­cal sup­port for these aes­thet­ic con­cepts, accord­ing to which the aes­thet­ic judge­ment is ahead of the analy­sis of the eval­u­at­ed work’s con­tent (Gołaszews­ka, 1984; Tatarkiewicz, 2009). On the oth­er hand, these data in no way ulti­mate­ly set­tle any­thing in the terms of beau­ty per­cep­tion. They do not give any for­mu­las for how to cre­ate works that will be liked, although they undoubt­ed­ly give some tips. These results are rather an illus­tra­tion of the research pos­si­bil­i­ties that ocu­log­ra­phy tech­niques open, the pur­pose of which is to record and analyse the eye move­ment while view­ing paint­ings. So let us remind and briefly com­ment on a few of them.

The time regard­ing casu­al view­ing paint­ings, i.e. with­out hav­ing to make val­ue judg­ments about them, was much longer than the time regard­ing view­ing them while mak­ing aes­thet­ic deci­sions. This rela­tion­ship seems quite obvi­ous, because it is a sim­ple con­se­quence of the instruc­tions, which before the phase I of the exper­i­ment encour­aged the sub­jects to view the col­lec­tion casu­al­ly and gave them full free­dom to decide on the time of view­ing each paint­ing. Accord­ing to the instruc­tions giv­en before the sec­ond phase of the exper­i­ment, the sub­jects were to aes­thet­i­cal­ly eval­u­ate the paint­ings which they had already viewed. In this phase, the sub­jects were not time-lim­it­ed either, but they did not need to devote as much time to this task as before, because they had already had suf­fi­cient visu­al knowl­edge regard­ing each painting.

Much more inter­est­ing is the result con­cern­ing the rela­tion­ship between the amount of time spent on view­ing paint­ings belong­ing to dif­fer­ent aes­thet­ic cat­e­gories in both phas­es of the exper­i­ment. Dur­ing the free-view­ing task, the sub­jects assigned high­er aes­thet­ic val­ues for the paint­ings they looked at much longer than those they looked at for a short­er time. Seem­ing­ly, it would seem that this result is con­sis­tent with the results of many oth­er exper­i­ments in which a sim­i­lar rela­tion­ship has been found (Var­tan­ian, Goel, 2004b). Nev­er­the­less, there is a fun­da­men­tal dif­fer­ence between them. It has tran­spired that while mak­ing an aes­thet­ic judg­ment, which was sep­a­rat­ed from aes­thet­ic expe­ri­ence (at least oper­a­tional­ly as part of this study), there are no dif­fer­ences regard­ing the time of view­ing paint­ings belong­ing to dif­fer­ent aes­thet­ic cat­e­gories. The deci­sion about belong­ing or non-belong­ing of a giv­en paint­ing to the cat­e­go­ry of beau­ti­ful paint­ings is tak­en by the respon­dents just as quickly.

To sum up, when we view paint­ings casu­al­ly, our gaze is focused for a longer time on those that we like more, but we make a judg­ment about them as quick­ly as in rela­tion to those paint­ings to which we pay much less atten­tion when view­ing. Such a dis­tinc­tion between expe­ri­ence and aes­thet­ic judg­ment is reflect­ed in the results of analy­ses on almost all para­me­ters of the eye move­ment tra­jec­to­ry, except for those that divide the time of view­ing paint­ings into sub­se­quent stages.

The results con­cern­ing the fix­a­tion dura­tion total on dif­fer­ent frag­ments of the viewed paint­ings reflect the results con­cern­ing their total view­ing time. First of all, the sub­jects much more focused their gaze on paint­ings viewed in phase I than in phase II. Sec­ond­ly, the fix­a­tion dura­tion total was much longer while casu­al view­ing paint­ings. The paint­ings belong to the most beau­ti­ful cat­e­gories than those that were rat­ed low­er due to aes­thet­ic rea­sons. Third­ly, while mak­ing aes­thet­ic judg­ments, the fix­a­tion dura­tion was sim­i­lar with respect to var­i­ous aes­thet­ic categories.

The fix­a­tion dura­tion total is an indi­ca­tor of gen­er­al inter­est in sen­so­ry data, i.e. the con­tent of those frag­ments of the viewed paint­ing that have inter­sect­ed the observer’s visu­al axes. There­fore, beau­ti­ful paint­ings aroused greater inter­est than less beau­ti­ful ones (Hauland, 2003; Latimer, 1988; Mel­lo-Thoms, Nodine and Kun­del, 2002).

The rela­tion­ships between the fix­a­tion dura­tion total also apply to the count of fix­a­tion points. There are many more of them in the I phase of the exper­i­ment than in the II one, more dur­ing free view­ing of paint­ings belong­ing to the high­er aes­thet­ic cat­e­go­ry than to the low­er one, and more or less the same in rela­tion to paint­ings belong­ing to dif­fer­ent aes­thet­ic cat­e­gories while eval­u­at­ing them in the II part of the experiment.

The count of fix­a­tion points, record­ed while view­ing the paint­ing, is one of deriv­a­tives of its view­ing time. How­ev­er, it is also an indi­ca­tor of the strat­e­gy of explor­ing the viewed paint­ing (Mas­saro, Savazzi, Di Dio, Freed­berg et al., 2012; Zange­meis­ter, Sher­man and Stark, 1995). Objects that are known, typ­i­cal or unin­ter­est­ing gen­er­al­ly do not require visu­al atten­tion for a longer time. Most often, they do not require ana­lyt­i­cal switch­ing of eyes from one image detail to anoth­er one. A much larg­er count of fix­a­tion points on var­i­ous frag­ments of paint­ings eval­u­at­ed to be par­tic­u­lar­ly beau­ti­ful may there­fore imply that they have some­thing unusu­al, unknown, and inter­est­ing. If there are more of them (in rela­tion to more beau­ti­ful paint­ings than less beau­ti­ful ones), it could mean that an ele­ment of nov­el­ty, sur­prise or the pres­ence of some detail in the image, which requires more atten­tion, is one of the com­po­nents of aes­thet­ic expe­ri­ence. Tak­ing as a basis the indi­ca­tors of fix­a­tion dura­tion total and fix­a­tion count, it is worth analysing two oth­er para­me­ters of the eye move­ment tra­jec­to­ry that cap­ture the ratio of dura­tion to their count. These are: fix­a­tion fre­quen­cy and fix­a­tion dura­tion average.

The fix­a­tion fre­quen­cy is the ratio of the count of fix­a­tion points to their fix­a­tion dura­tion total, while the fix­a­tion dura­tion aver­age is the ratio of the fix­a­tion dura­tion total to their count. There­fore, both indi­ca­tors imply more or less the same phe­nom­e­non, i.e. the observer’s aver­age involve­ment in pro­cess­ing per­cep­tu­al data, aver­aged for all fix­a­tion points (Mas­saro, Savazzi, Di Dio, Freed­berg et al., 2012).

The fix­a­tion dura­tion aver­age is an indi­ca­tor of the strength of per­cep­tu­al data pro­cess­ing (Duchows­ki, 2007; Mol­nar, 1981). When focus­ing on one point of the image, not only sen­so­ry data is record­ed using recep­tors locat­ed in the reti­na, but they are also trans­mit­ted to the cor­ti­cal struc­tures of the brain and sub­ject­ed to com­plex ana­lyt­i­cal process­es, which results in the deci­sion to move the eye­ball to anoth­er place of the image. These process­es gen­er­al­ly take place out­side con­scious­ness and often beyond the observer’s will as well. The hard­er to inter­pret the sen­so­ry data, the longer it is analysed. These mil­lisec­ond times for per­for­mance of indi­vid­ual neu­ro­phys­i­o­log­i­cal and cog­ni­tive activ­i­ties aggre­gate, deter­min­ing the total dura­tion of sin­gle fix­a­tion. A longer fix­a­tion dura­tion aver­age on a spe­cif­ic frag­ment of a giv­en paint­ing most like­ly indi­cates a greater cog­ni­tive effort relat­ed to its inter­pre­ta­tion. It is the result of high­er require­ments for cog­ni­tive data pro­cess­ing that is cur­rent­ly in the field of vision (Rayn­er, 1998). This would imply that more beau­ti­ful paint­ings (in con­trary to less beau­ti­ful ones) con­tain more ambigu­ous ele­ments that require more time to watch them.

The results of the analy­sis on the fix­a­tion dura­tion aver­ages and fix­a­tion fre­quen­cies part­ly con­firm these assump­tions. It tran­spires that while the val­ues ​​of both indi­ca­tors are actu­al­ly longer when casu­al view­ing paint­ings than while mak­ing aes­thet­ic judg­ments about them, they do not dif­fer – both in phas­es I and II of the exper­i­ment – depend­ing on the aes­thet­ic cat­e­go­ry of paint­ings. These results imply that while aes­thet­ic expe­ri­ence is accom­pa­nied by a deep­er, cog­ni­tive analy­sis of sen­so­ry data, there is a quick deci­sion-mak­ing process (refer­ring to knowl­edge formed dur­ing ear­li­er view­ing of a paint­ing) dur­ing the aes­thet­ic judg­ment. It can­not be exclud­ed that the sub­ject of the eval­u­a­tion is not so much the scope of con­tent of the works viewed, but rather the amount of cog­ni­tive effort put into their inter­pre­ta­tion. How­ev­er, this rela­tion­ship seems to be much more com­plex. It is dif­fi­cult to agree with the fact that the more cog­ni­tive effort the observ­er puts into the cog­ni­tive inter­pre­ta­tion of a giv­en paint­ing, the more (s)he likes it. It is pos­si­ble that an impor­tant fac­tor influ­enc­ing this rela­tion­ship is allow­ing the sub­jects to choose the amount of time they want to spend on view­ing a work. How­ev­er, there are prob­a­bly some ranges of this involve­ment that are opti­mal from the point of view of aes­thet­ic eval­u­a­tion of the work.

Sum­maris­ing the cur­rent dis­cus­sion con­cern­ing the fix­a­tion points on viewed paint­ings, two phe­nom­e­na can be formulated.

First­ly, accord­ing to, adopt­ed in the present research, oper­a­tional­i­sa­tion of aes­thet­ic expe­ri­ence and aes­thet­ic judg­ment for phase I and II of the exper­i­ment, it can be stat­ed that they are man­aged by var­i­ous neu­ro­phys­i­o­log­i­cal sys­tems. These sys­tems are respon­si­ble for the visu­al explo­ration of the per­cep­tu­al field and the strength of sen­so­ry data pro­cess­ing. Aes­thet­ic expe­ri­ence is char­ac­terised by a much greater explo­ration of the per­cep­tu­al field and more ana­lyt­i­cal, as well as more involved, pro­cess­ing than aes­thet­ic eval­u­a­tion. Metaphor­i­cal­ly speak­ing, the mak­ing of an aes­thet­ic judg­ment does not require a “return to the image” but it is based on inter­nalised assump­tions formed dur­ing the aes­thet­ic experience.

Sec­ond­ly, although it has tran­spired that (in rela­tion to paint­ings that were rat­ed high­er because of their aes­thet­ic val­ue) the fix­a­tion dura­tion total and count of fix­a­tion points increase while casu­al view­ing paint­ings, it is pos­si­ble that this effect is caused by the glob­al length of time spent on analysing paint­ings in the part I of the experiment.

The sec­ond group of indi­ca­tors for eye move­ment tra­jec­to­ry is a saccade.

The total scan­path length is equal to the length of the path which the eye fol­lows on the plane of the paint­ing dur­ing its expo­sure. Sim­i­lar­ly to the total fix­a­tion dura­tion indi­ca­tor, the length of the scan­ning path of a giv­en paint­ing is also one of deriv­a­tives of its view­ing time. There­fore, since the amount of time spent on view­ing paint­ings in part I of the exper­i­ment was sig­nif­i­cant­ly longer than the amount of time spent on mak­ing an aes­thet­ic judg­ment, it is not sur­pris­ing that the scan­ning paths for paint­ings dur­ing their free view­ing were much longer than dur­ing view­ing them just before the aes­thet­ic evaluation.

Refer­ring to glob­al time spent on view­ing paint­ings, one can also explain the effect of a sig­nif­i­cant­ly longer scan­ning path for paint­ings with high­er aes­thet­ic val­ue, in com­par­i­son to the cat­e­go­ry of paint­ings with low­er aes­thet­ic val­ue, as part of the col­lec­tion in ques­tion. Obvi­ous­ly, this is only about the phase I of the exper­i­ment, as the dif­fer­ences between the total scan­path lengths, when view­ing paint­ings belong­ing to dif­fer­ent aes­thet­ic cat­e­gories, were very sim­i­lar in the phase II. One can find out if the aes­thet­ic cat­e­go­ry has an impact on the sac­cade length by esti­mat­ing the indi­ca­tor of aver­age sac­cade length, i.e. the quo­tient of the total scan­ning path by the sac­cade number.

The analy­sis of the aver­age sac­cade length has revealed inter­est­ing rela­tion­ships. It has tran­spired that although total scan­ning paths of the paint­ings dur­ing the phase I of the exper­i­ment were much longer than dur­ing the phase II, the aver­age sac­cade length was almost iden­ti­cal in both stages of the exper­i­ment. This means that due to this indi­ca­tor, expe­ri­ence and aes­thet­ic judg­ment do not dif­fer from each oth­er. It has turned out that dur­ing both the aes­thet­ic expe­ri­ence and the aes­thet­ic judg­ment, sac­cades are sig­nif­i­cant­ly short­er in rela­tion to paint­ings with high­er aes­thet­ic val­ues, ​​than in rela­tion to paint­ings with low­er aes­thet­ic values.

The aver­age sac­cade length is inter­pret­ed in terms of the cog­ni­tive strat­e­gy adopt­ed by the observ­er (Humprey and Under­wood, 2009; Mas­saro, Savazzi, Di Dio, Freed­berg et al., 2012; Zange­meis­ter, Sher­man and Stark, 1995). Long sac­cades, or eye move­ment to dis­tant frag­ments of the image, rather indi­cate a strat­e­gy of glob­al search­ing the per­cep­tu­al field in order to build some­thing like a cog­ni­tive map of the image. Ini­tial phas­es of view­ing an image are fre­quent­ly char­ac­terised by long sac­cades. In par­tic­u­lar, this applies to scenes that con­tain new objects or rela­tions between them that are not always under­stood at first glance. Long sac­cades allow cog­ni­tive recon­struc­tion of the image com­po­si­tion. In turn, short sac­cades indi­cate a more local analy­sis of a small frag­ment of the paint­ing, which requires more detailed explo­ration for some rea­son. They may be a man­i­fes­ta­tion of dif­fi­cul­ties in recog­nis­ing an object in the scene or an expres­sion of the need to decide on the mean­ings con­tained in a giv­en fragment.

The peo­ple sur­veyed in the pre­sent­ed exper­i­ment belonged to a group of lay­men in the field of art, there­fore it can be assumed that they used more bot­tom-up than top-down, and rather local than glob­al, strate­gies. On the basis of the avail­able data, how­ev­er, it can­not be deter­mined whether they used top-down or bot­tom-up strat­e­gy, but the fact is that for more beau­ti­ful paint­ings they more often per­formed short­er sac­cades than for paint­ings which were rat­ed low­er due to their aes­thet­ic val­ue. On the one hand, this may mean that they were sim­ply more ana­lyt­i­cal­ly involved in solv­ing some local prob­lems relat­ed to their inter­pre­ta­tion. This is a typ­i­cal behav­iour of novices in the field to which visu­al scene relates. On the oth­er hand, longer ampli­tudes of sac­cades when view­ing paint­ings belong­ing to low­er aes­thet­ic cat­e­gories would rather indi­cate behav­iour typ­i­cal of experts. After all, experts are more like­ly to use glob­al strate­gies than lay­men (Zange­meis­ter, Sher­man and Stark, 1995).

The jux­ta­po­si­tion of these two points of view sheds an inter­est­ing light on the rela­tion­ship that occurs among the eye­ball sac­cades, being an expert or a lay­man in the field of art and the aes­thet­ic eval­u­a­tion of a giv­en work. The lay­men sur­veyed in the field of art viewed less beau­ti­ful paint­ings in accor­dance with glob­al strat­e­gy, i.e. they behaved like experts. How­ev­er, they behaved as novices while view­ing beau­ti­ful paint­ings, using rather local strate­gies of search­ing the per­cep­tu­al field. Per­haps the paint­ings to which the sub­jects attrib­uted low­er aes­thet­ic grades are sim­ply more typ­i­cal and pre­dictable. In more beau­ti­ful paint­ings, they dis­cov­ered visu­al rid­dles in turn, which raised cog­ni­tive anx­i­ety in them and held their visu­al atten­tion on some frag­ments longer.

Being an expert in a giv­en field is tan­ta­mount to knowl­edge of objects belong­ing to this field. If so, the expert gen­er­al­ly does not car­ry out a detailed analy­sis of the famil­iar object and rather uses a glob­al strat­e­gy of search­ing the per­cep­tu­al field.  Simul­ta­ne­ous­ly, the accom­pa­ny­ing sense of famil­iar­i­ty with spe­cif­ic objects (e.g. paint­ings) makes them less inter­est­ing, less mys­te­ri­ous, and more banal. From here, it is only one step to their low­er aes­thet­ic eval­u­a­tion. For a novice in turn, many objects belong­ing to an unknown field may seem mys­te­ri­ous, intrigu­ing and involv­ing. Symp­toms of these sit­u­a­tions are vis­i­ble in the eye move­ment tra­jec­to­ry record­ed when the exam­ined paint­ings of cat­e­go­ry A were viewed: short­er ampli­tudes of sac­cades, longer time spent on view­ing them, and a greater fix­a­tion count. This inter­pre­ta­tion of the exper­i­men­t’s results is char­ac­terised by the fol­low­ing para­dox: the more some­one sees them­selves as an expert in the field of art, the more dif­fi­cult is to see beau­ty in the paint­ings viewed.

Cer­tain­ly, the results of the pre­sent­ed stud­ies do not explain the aes­thet­ic sen­sa­tions in a ful­ly sat­is­fac­to­ry way, using their ocu­lo­mo­tor indi­ca­tors. Instead, they reveal some inter­est­ing clues that can (and even should) become the sub­ject of a fur­ther exper­i­men­tal study. First of all, one should exam­ine to what extent the rela­tion­ships pre­sent­ed here relate to oth­er types of paint­ings, e.g. land­scapes or abstract works. Undoubt­ed­ly, the study of ocu­lo­mo­tor activ­i­ty seems to be the most nat­ur­al form of col­lect­ing data on the behav­iour of peo­ple view­ing works of visu­al art. The pre­sent­ed analy­sis con­firms this intu­ition completely.

Table of content