Function ofthethalamic reticular complex: Thesearchlight

Itissuggested thatinthebrain theinternal attentional searchlight, proposed byTreisman andothers, is controlled bythereticular complex ofthethalamus (including theclosely related perigeniculate nucleus) andthat theexpres- sion ofthesearchlight istheproduction ofrapid bursts offir- inginasubset ofthalamic neurons. Itisalso suggested that the conjunctions produced bytheattentional searchlight aremedi- atedbyrapidly modifiable synapses herecalled Malsburg synapses-and especially byrapid bursts acting onthem. The activation ofMalsburg synapses isenvisaged asproducing transient cell assemblies, including "vertical" onesthat tempo- rarily unite neurons atdifferent levels intheneural hierarchy. This paper presents asetofspeculative hypotheses concern- ingthefunctions ofthethalamus and,inparticular, thenu- cleus reticularis ofthethalamus and, therelated perigenicu- late nucleus. Foreaseofexposition Ihavedrawnmyexam- ples mainly fromthevisual system ofprimates, butIexpect theideas toapply toall mammalsandalso toother systems, suchasthelanguage system inman. Visual System Itisnowwellestablished thatintheearly visual system of primates there areatleast 10distinct visual areasintheneo- cortex. (For arecent summary, seeVanEssenandMaunsell (1).) Ifweinclude allareaswhosemainconcern iswithvi- sion, there maybeperhaps twice thatnumber. Toagood approximation, theearly visual areascanbearranged ina branching hierarchy. Eachofthese areas hasacrude"map" of(part of)thevisual world. Thefirst visual area(area 17, also called thestriate cortex) ononeside oftheheadmaps one-half ofthevisual world inrather fine detail. Itscells can respond torelatively simple visual "features," suchasorien- tation, spatial frequency, disparity (between thetwoeyes), etc. Thisparticular areaisalarge onesothat theconnections between different parts ofitarerelatively local. Eachpart therefore responds mainly totheproperties ofasmall local partofthevisual field (2). Asoneproceeds toareashigher inthehierarchy, the "mapping" becomes morediffuse. Atthesametimetheneu- ronsappear torespond tomorecomplex features inthevisu- alfield. Different cortical areas specialize, tosomeextent, in different features, oneresponding mainly tomotion, another moretocolor, etc. Inthehigher areas aneuron hardly knows whereinthevisual field thestimulus (such asaface) isaris- ing, while thefeature itresponds tomaybesocomplex that individual neurons areoften difficult tocharacterize effec- tively (3,4). Thus, thedifferent areas analyze thevisual field indiffer- entways.Thisisnot,however, howweappear toseethe world. Ourinner visual picture oftheexternal worldhasa unity. Howthendoesthebrain puttogether allofthese dif- ferent activities toproduce aunified picture sothat, forex- ample, foranyobject theright color isassociated withthe right shape? TheSearchlight Thepioneer workofTreisman andhercolleagues (5-8), sup- ported morerecently bytheelegant experiments ofJulesz (9-11), haverevealed aremarkable fact. Ifonlyaveryshort space oftimeisavailable, especially inthepresence of"dis- tractors," thebrain isunable tomakethese conjunctions re- liably. Forexample, ahumansubject canrapidly spot an"S" mixedinwitha randomly arranged setofgreenXsand brownTs-it"popsout"athim.Hisperformance isalso rapid forablueletter mixedinwiththesameset. However, ifheisaskedtodetect agreenT(which requires thathe recognize theconjunction ofachosen color withachosen shape), heusually takesmuchmoretime.Moreover, the timeneeded increases linearly withthenumberofdistractors (the green XsandbrownTs)asifthemindweresearching theletters inseries, asifthebrain hadaninternal attentional searchlight that movedaround fromonevisual object tothe next, withsteps asfast as70msecinfavorable cases. Inthis metaphor thesearchlight isnotsupposed tolight uppart ofa completely darklandscape but, like asearchlight atdusk, it intensifies part ofascene that isalready visible tosomeex- tent. Ifthere isindeed asearchlight mechanism inthebrain, howdoesitworkandwhereisitlocated? Toapproach this problem wemuststudy thegeneral layout ofthebrain and, inparticular, thatoftheneocortex andthethalamus. The essential facts weneedatthis stage areasfollows.