Sacrificing Accuracy to See the Big Picture

Our ability to detect patterns might stem, in part, from the brain’s desire to represent things in the simplest way possible, according to University of Pennsylvania researchers.


  • newswise-fullscreen Sacrificing Accuracy to See the Big Picture

    Credit: Lynn et al.

    Subjects respond to sequences of stimuli drawn as a random walk on an underlying transition graph. A. Example sequence of visual stimuli (left) representing a random walk on an underlying transition network (right). B. For each stimulus, subjects are asked to respond by pressing a combination of one or two buttons on a keyboard. C. Each of the 15 possible button combinations corresponds to a node in the transition network. We only consider networks with nodes of uniform degree k = 4 and edges with uniform transition probability 0.25. D. Subjects were asked to respond to sequences of 1500 such nodes drawn from two different transition architectures: a modular graph (left) and a lattice graph (right). E. Average reaction times across all subjects for the different button combinations, where the diagonal elements rep- resent single-button presses and the off-diagonal elements represent two-button presses. F. Average reaction times as a function of trial number, characterized by a steep drop-off in the first 500 trials followed by a gradual decline in the remaining 1,000 trials.

Newswise — WASHINGTON, D.C., March 5, 2019 -- During their first year of life, infants can recognize patterned sound sequences. As we grow, we develop the ability to pick out increasingly complex patterns within streams of words and musical notes. Traditionally, cognitive scientists have assumed that the brain uses a complicated algorithm to find links between disparate concepts, thereby yielding a higher-level understanding. 

Researchers at the University of Pennsylvania -- Christopher Lynn, Ari Kahn and Danielle Bassett -- are building an entirely different model, indicating that our ability to detect patterns might stem, in part, from the brain’s desire to represent things in the simplest way possible. 

The brain does more than just process incoming information, said Lynn, a physics graduate student. “It constantly tries to predict what’s coming next. If, for instance, you’re attending a lecture on a subject you know something about, you already have some grasp of the higher-order structure. That helps you connect ideas together and anticipate what you’ll hear next.” 

The new model offers striking insights on human cognition, suggesting that people can and indeed do make mistakes in detecting individual components of a pattern in order to catch a glimpse of the bigger picture. For example, Lynn explained, “if you look at a pointillist painting up close, you can correctly identify every dot. If you step back 20 feet, the details get fuzzy, but you’ll gain a better sense of the overall structure.” The brain may well adopt a similar strategy, he said. 

To test its theory, the team developed an experiment in which people view a computer screen depicting a row of five squares and then press one or two keys to match the display they see. The researchers timed the responses, concluding that people hit the correct keys more quickly when they anticipate what’s coming next. 

As part of the experimental setup, each stimulus presented to a subject could be viewed as a node in a network, with one of four adjacent nodes representing the next stimulus. The networks come in two forms -- a “modular graph” consisting of three linked pentagons and a “lattice graph” consisting of five linked triangles. Subjects reacted more quickly when presented with the modular graph, suggesting they could better discern its underlying structure and, hence, better anticipate the image to follow. 

Ultimately, the experiment is designed to measure a quantity the authors call beta (β), which varies from subject to subject, assuming lower values in people prone to making errors and higher values in those inclined toward accuracy. The Pennsylvania group plans to acquire brain images from fMRI scans later this year to see if the brains of people found to have different values of β are, in fact, “wired differently.” 

### 

The 2019 APS March Meeting presentation “Structure from noise: Mental errors yield abstract representations of events," by Christopher Lynn, Ari E. Kahn and Danielle Bassett, will take place Tuesday, March 5, at 3:06 p.m. in room 261 of the Boston Convention and Exhibition Center. ABSTRACT: https://meetings.aps.org/Meeting/MAR19/Session/H66.4 

----------------------- MORE MEETING INFORMATION ----------------------- 

The American Physical Society (APS) March Meeting is a major international conference and the largest physics meeting of the year. In 2019, the APS March Meeting will convene March 4-8 at the Boston Convention and Exhibition Center. 

USEFUL LINKS 

Register as Press: https://goo.gl/forms/ur9dE24zCO1IVsLu2 
Press Site: https://www.aps.org/meetings/march/press.cfm 
Meeting Abstracts: http://meetings.aps.org/Meeting/MAR19/APS_epitome 
Main Meeting Page: https://www.aps.org/meetings/march/ 
Hotel and Travel: https://www.aps.org/meetings/march/hotel-travel.cfm 

PRESS REGISTRATION 

APS will provide free registration to all staff journalists representing media organizations, professional freelance journalists on assignment, and student journalists who are attending the meeting for the express purpose of gathering and reporting news and information. Press registration grants full access to all scientific sessions, to the press room, and to the press conferences. We will also provide complimentary press registration to university press officers, PIOs and other professional media relations staff. For press related questions about the APS March Meeting, email media@aps.org

PRESS CONFERENCES 

A series of press conferences on newsworthy research will be webcast live from the conference on Monday, March 4 through Thursday, March 7. A schedule will be released the week before the conference. Members of the media should register in advance at https://webcast.apswebcasting.com/go/aps-march-19

ABOUT APS 

The American Physical Society is a nonprofit membership organization working to advance and diffuse the knowledge of physics through its outstanding research journals, scientific meetings, and education, outreach, advocacy, and international activities. APS represents over 55,000 members, including physicists in academia, national laboratories, and industry in the United States and throughout the world. Society offices are located in College Park, Maryland (Headquarters), Ridge, New York, and Washington, D.C. Read more: https://www.aps.org/ 

###

  • share-facebook-Sacrificing Accuracy to See the Big Picture
  • share-twitter-Sacrificing Accuracy to See the Big Picture

Comment/Share

Chat now!