Study: Brain imaging, combined with computer modeling, can help predict your emotional responses

Source: Occip­i­tal-tem­po­ral cor­ti­cal tun­ing to seman­tic and affec­tive fea­tures of nat­ur­al images pre­dicts asso­ci­at­ed behav­ioral respons­es (2024)

In a new study, researchers were able to pre­dict a person’s response to emo­tion­al­ly charged scenes using brain imag­ing and com­put­er mod­el­ing alone—gauging not only whether the person’s reac­tion was pos­i­tive, neg­a­tive, or neu­tral, but also how strong the reac­tion was.

The study helps neu­ro­sci­en­tists under­stand how the brain rep­re­sents com­plex emo­tion­al nat­ur­al stim­uli, accord­ing to senior author Sonia Bish­op, adjunct asso­ciate pro­fes­sor of neu­ro­science at the Uni­ver­si­ty of Cal­i­for­nia, Berke­ley, and the new­ly appoint­ed chair of psy­chol­o­gy at Trin­i­ty Col­lege Dublin.

The sim­ple tasks used in the research will also make it eas­i­er to study autism spec­trum dis­or­der, where researchers seek to under­stand how indi­vid­u­als dif­fer in pro­cess­ing every­day emo­tion­al stimuli.

The study by neu­ro­sci­en­tists at UC Berke­ley, Trin­i­ty Col­lege Dublin, and Google was pub­lished in July in the jour­nal Nature Communications.

“It is huge­ly impor­tant for all species to be able to rec­og­nize and respond appro­pri­ate­ly to emo­tion­al­ly salient stim­uli, whether that means not eat­ing rot­ten food, run­ning from a bear, approach­ing an attrac­tive per­son in a bar, or com­fort­ing a tear­ful child,” said Bish­op, who is also a mem­ber of UC Berkeley’s Helen Wills Neu­ro­science Insti­tute. “How the brain enables us to respond in a nuanced way to emo­tion­al­ly charged sit­u­a­tions and stim­uli has long been of inter­est, but lit­tle is known about how the brain stores schemas or neur­al rep­re­sen­ta­tions to sup­port the nuanced behav­ioral choic­es we make in response to emo­tion­al nat­ur­al stimuli.”

In addi­tion, few stud­ies have looked beyond a sim­ple bina­ry reaction—approach or avoid, fight or flight—when humans clear­ly have a more nuanced response.

“Neu­ro­science stud­ies of moti­vat­ed behav­ior often focus on sim­ple approach or avoid­ance behav­iors, such as lever-press­ing for food or chang­ing loca­tions to avoid a shock,” she said. “How­ev­er, when faced with nat­ur­al emo­tion­al stim­uli, humans don’t sim­ply choose between ‘approach’ or ‘avoid.’ Rather, they select from a com­plex range of suit­able respons­es. So, for exam­ple, our avoid response to a large bear—leave the area ASAP—is dif­fer­ent to our avoid response to a weak, dis­eased animal—don’t get too close. Sim­i­lar­ly, our approach response to the pos­i­tive stim­uli of a poten­tial mate dif­fers from our approach reac­tion to a cute baby.”

In the new study, led by for­mer UC Berke­ley doc­tor­al stu­dent Samy Abdel-Ghaf­far, who is now at Google, human vol­un­teers were shown a vari­ety of nat­ur­al images—a baby’s face, a snarling dog, a per­son vomiting—chosen to evoke an emo­tion­al response. The par­tic­i­pants’ 3D brain activ­i­ty was mea­sured with a func­tion­al mag­net­ic res­o­nance imager (fMRI); they also were asked to rate the images as pos­i­tive, neg­a­tive, or neu­tral and report­ed the degree of emo­tion­al arousal to each.

Analy­sis of brain-wide activ­i­ty showed that regions of the occip­i­tal tem­po­ral cor­tex, locat­ed in the back of the brain, are tuned to rep­re­sent both the type of stimulus—single human, cou­ple, crowd, rep­tile, mam­mal, food, object, build­ing, landscape—and the emo­tion­al char­ac­ter­is­tics of the stim­u­lus. For exam­ple, pos­i­tive high-arousal faces were rep­re­sent­ed in slight­ly dif­fer­ent regions than neg­a­tive high-arousal faces or neu­tral low-arousal faces.

“Our research reveals that the occip­i­tal tem­po­ral cor­tex is tuned not only to dif­fer­ent cat­e­gories of stim­uli; it also breaks down these cat­e­gories based on their emo­tion­al char­ac­ter­is­tics in a way that is well suit­ed to guide selec­tion between alter­nate behav­iors,” Bish­op said.

Abdel-Ghaf­far then used machine learn­ing, a type of arti­fi­cial intel­li­gence, to pre­dict the response of a sec­ond group of vol­un­teers to the same images based sole­ly on the sta­ble tun­ing pat­terns in the occip­i­tal tem­po­ral cor­tex. He found that he could. In fact, ana­lyz­ing brain activ­i­ty was a bet­ter pre­dic­tor of par­tic­i­pants’ reac­tions than a machine learn­ing assess­ment of the emo­tion­al aspects of the actu­al images.

“This sug­gests that the brain choos­es which infor­ma­tion is impor­tant or not impor­tant to rep­re­sent and holds sta­ble rep­re­sen­ta­tions of sub-cat­e­gories of ani­mate and inan­i­mate stim­uli that inte­grate affec­tive infor­ma­tion and are opti­mal­ly orga­nized to sup­port the selec­tion of behav­iors to dif­fer­ent types of emo­tion­al nat­ur­al stim­uli,” Bish­op said.

She not­ed also that “the par­a­digm used does not involve a com­plex task, mak­ing this approach suit­able in the future, for exam­ple, to fur­ther under­stand­ing of how indi­vid­u­als with a range of neu­ro­log­i­cal and psy­chi­atric con­di­tions dif­fer in pro­cess­ing emo­tion­al nat­ur­al stimuli.”

– Robert Sanders is man­ag­er of sci­ence com­mu­ni­ca­tions at UC Berke­ley and writes about fields rang­ing from biotech­nol­o­gy to brain research. This arti­cle was orig­i­nal­ly pub­lished on Berke­ley News. Read the orig­i­nal arti­cle and the source study.

To Learn More:

• Study: Prac­tic­ing grat­i­tude helps low­er phys­i­o­log­i­cal stress lev­els of the express­er AND the receiver
• Explor­ing the human brain and how it responds to stress