(consciousness redux) Probing the Unconscious Mind Cognitive psychology is mapping the capabilities we are unaware we possess BY CHRISTOF KOCH SIGMUND FREUD popular- Van Opstal and his col- ized the idea of the uncon- leagues proved the oppo- scious, a sector of the mind site in an indirect but clever that harbors thoughts and and powerful way. memories actively removed A quartet of single- from conscious deliberation. digit Arabic numbers (1 Because this aspect of mind through 9, excluding the is, by definition, not accessi- numeral 5) are projected ble to introspection, it has onto a screen. Volunteers proved difficult to investi- had to indicate as quickly gate. Today the domain of as possible whether or not the unconscious—described the average of the four pro- more generally in the realm jected numbers exceeded of cognitive neuroscience as 5. Every trial was preceded any processing that does not by a hidden cue that could give rise to conscious aware- be valid or invalid. The cue ness—is routinely studied in consisted of a very brief hundreds of laboratories us- flash of another set of four ing objective psychophysical numbers whose average techniques amenable to sta- was either smaller or larg- tistical analysis. Let me tell er than 5 [see illustration you about two experiments below]. These were pre- that reveal some of the capabilities of the lands and Stanislas Dehaene of the Col- ceded and followed by hash marks at the unconscious mind. Both depend on lège de France in Paris. Dehaene, director location of the flashed numbers. The “masking,” as it is called in the jargon, of the INSERM-CEA Cognitive Neuro- marks effectively masked the cue so that ) or hiding things from view. Subjects imaging Unit, is best known for his inves- no subject ever consciously saw this quar- look but don’t see. tigations of the brain mechanisms under- tet. Forcing them to guess whether the av- lying counting and numbers. Here he ex- erage of the four hidden numbers was less Unconscious Arithmetic plored the extent to which a simple sum than or greater than 5 did not work ei- The first experiment is a collaboration or an average can be computed outside the ther: they were at chance. among Filip Van Opstal of Ghent Univer- pale of consciousness. Adding 7, 3, 5 and Yet the cue still influenced the sub- woman in clothes hanger clothes in woman sity in Belgium, Floris P. de Lange of Rad- 8 is widely assumed to be a quintessential ject’s reaction to the main response. If the ( boud University Nijmegen in the Nether- serial process that requires consciousness. implicit cue was valid, the response to the target was consistently faster than if the cue was invalid. In the illustration, the Getty Images In the experiment, subjects saw four numbers for 600 milliseconds and had to judge quickly whether their average exceeded 5. Masks with hash marks ensured that the four cued numbers mean of the four invisible cues (3.75) is M were not consciously seen. The unconscious was nonetheless able to estimate the average. less than 5, whereas the average of the vis- LIA. U ible target numbers is greater than 5. Re- ); Y Start Mask Cue Mask Target solving this conflict demands additional 1 Koch processing time (about ⁄40 of a second). # 2 # 8 + #+ # 6 +1 # + # 7+ 3 That is, the cue triggers neural activity # 6 # 6 representing the assertion “less than 5,” which interferes with the rapid establish- HRISTOF KOCH ( KOCH HRISTOF Time ment of a coalition of neurons represent- C 20 SCIENTIFIC AMERICAN MIND November/December 2011 The ability to rapidly integrate disparate elements in a scene ( and place them in context is a hallmark of consciousness. ) ing “greater than 5.” That invisible and these pictures: a woman puts a chess- undetectable cues influence behavior im- board into the oven, the cocked arrow is plies that the unconscious can somehow replaced by a tennis racket, and the bas- estimate the average of four single digits. ketball becomes a watermelon. The psy- It is unlikely that it does so following the chologists made sure that both congru- precise, algebraic rules children learn in ent and incongruent images were truly grade school. Instead it may rely on heu- invisible and could not be distinguished ristics: for example, for each number from one another when masked in this larger than 5, increase the probability of way. This discovery implies that the un- pushing the greater than 5 button. conscious can recognize something is This is just the latest in a flurry of ex- amiss in these images, that the object periments demonstrating so-called en- handled by the person in the image is not semble coding, the ability of the mind to appropriate to the context. guesstimate the dominant emotional ex- How the mind recognizes that some- pression of a crowd of faces or the ap- thing is wrong is puzzling. Maybe be- proximate size of a bunch of dots even cause the vast and tangled neural net- though the individual faces or dots are works of the cerebral cortex that encode not consciously perceived. images have learned that certain objects go together but others do not (akin to the What’s Wrong with this Picture? software programs—bots—that Google Liad Mudrik and Dominique Lamy and other search engines employ to trawl of Tel Aviv University and Assaf Breska the Internet to list all images, sentences and Leon Y. Deouell of the Hebrew Uni- and Web pages so when you search for versity in Jerusalem set out to test the ex- them they are readily accessible). Given tent to which the unconscious can inte- the sheer infinite number of possible grate all the information in any one pic- pairings of objects and context, is this so- ture into a unified and coherent visual lution likely to be done by the brain? Or experience. Giulio Tononi and I had pro- The unconscious mind can tell if there is maybe the masking techniques suppress posed in the last Consciousness Redux something amiss in these doctored images. visibility of the image but do not fully column [September/October 2011] that eliminate conscious access to them? the ability to rapidly integrate all the dis- The fascinating aspect of the Mudrik Only more research will tell. In this way, parate elements within a scene and place study is that the time to become visible we shall ultimately know the capabilities them into context is one of the hall- depends on the content of the image. Re- of the cognitive unconscious and the tru- marks of consciousness. alistic scenes that depict a woman plac- ly essential function that consciousness The Israeli researchers used “continu- ing a pizza into an oven, a boy taking aim plays in our life. M ous flash suppression,” a powerful mask- with a bow and arrow, or a basketball ing technique, to render images invisible. player dunking a ball into a hoop took CHRISTOF KOCH is Lois and Victor Troendle A series of rapidly changing, randomly 2.64 seconds to become visible, whereas Professor of Cognitive and Behavioral Biology colored patterns was flashed into one eye unnatural scenes were masked for only at the California Institute of Technology and while a photograph of a person carrying 2.50 seconds, a small but significant dif- chief scientific officer at the Allen institute for out some task was slowly faded into the ference. That is, the unconscious mind Brain Science in Seattle. He serves on Scien- Tel AvivTel University other eye. For a few seconds, the picture detected something incongruent about tific American Mind’s board of advisers. DRIK is completely invisible, and the subject can Mu see only the colored shapes. Because the IAD images become progressively stronger, (Further Reading) L eventually they will break through, and ◆ Integration without Awareness: Expanding the Limits of Unconscious Processing. OF Y the subject will see them. It is like Harry Liad Mudrik, Assaf Breska, Dominique Lamy and Leon Y. Deouell in Psychological Science, Vol. 22, No. 6, pages 764–770; June 2011. RTES Potter’s cloak of invisibility fading with U ◆ Rapid Parallel Semantic Processing of Numbers without Awareness. Filip Van Opstal, Floris P. O C time and revealing what is underneath. de Lange and Stanislas Dehaene in Cognition, Vol. 120, No. 1, pages 136–147; July 2011. www.ScientificAmerican.com/Mind SCIENTIFIC AMERICAN MIND 21.