Integration of What and Where in the Primate Prefrontal Cortex
10.1126/science.276.5313.821
Crossref journal-article
American Association for the Advancement of Science (AAAS)
Science (221)
Abstract

The visual system separates processing of an object’s form and color (“what”) from its spatial location (“where”). In order to direct action to objects, the identity and location of those objects must somehow be integrated. To examine whether this process occurs within the prefrontal (PF) cortex, the activity of 195 PF neurons was recorded during a task that engaged both what and where working memory. Some neurons showed either object-tuned (what) or location-tuned (where) delay activity. However, over half (52 percent, or 64/123) of the PF neurons with delay activity showed both what and where tuning. These neurons may contribute to the linking of object information with the spatial information needed to guide behavior.

Bibliography

Rao, S. C., Rainer, G., & Miller, E. K. (1997). Integration of What and Where in the Primate Prefrontal Cortex. Science, 276(5313), 821–824.

Authors 3
  1. S. Chenchal Rao (first)
  2. Gregor Rainer (additional)
  3. Earl K. Miller (additional)
References 35 Referenced 627
  1. Reviewed in L. G. Ungerleider and M. Mishkin in Analysis of Visual Behavior M. A.Goodale and R. J. W. Mansfield Eds. (MIT Press Cambridge MA 1982) pp. 549–586;
  2. Maunsell J. H. R., Newsome W. T., Annu. Rev. Neurosci. 10, 363 (1987). (10.1146/annurev.ne.10.030187.002051) / Annu. Rev. Neurosci. by Maunsell J. H. R. (1987)
  3. D. N. Pandya and C. L. Barnes in The Frontal Lobes Revisited E. Perecman Ed. (IRBN New York 1987) pp. 41–72.
  4. A. Baddeley Working Memory (Clarendon Oxford 1986).
  5. Fuster J. M., Alexander G. E., Science 173, 652 (1971); (10.1126/science.173.3997.652) / Science by Fuster J. M. (1971)
  6. Kubota K., Niki H., J. Neurophysiol. 34, 337 (1971); (10.1152/jn.1971.34.3.337) / J. Neurophysiol. by Kubota K. (1971)
  7. Fuster J. M., ibid. 36, 61 (1973); / ibid. by Fuster J. M. (1973)
  8. ; S. Funahashi C. J. Bruce P. S. Goldman-Rakic ibid. 61 331 (1989); (10.1152/jn.1989.61.2.331)
  9. Miller E. K., Erickson C. A., Desimone R., J. Neurosci. 16, 5154 (1996). (10.1523/JNEUROSCI.16-16-05154.1996) / J. Neurosci. by Miller E. K. (1996)
  10. Petrides M., Pandya D. N., J. Comp. Neurol. 228, 105 (1984); (10.1002/cne.902280110) / J. Comp. Neurol. by Petrides M. (1984)
  11. Ungerleider L. G., Gaffan D., Pelak V. S., Exp. Brain Res. 76, 473 (1989). (10.1007/BF00248903) / Exp. Brain Res. by Ungerleider L. G. (1989)
  12. Barbas H., Pandya D. N., J. Comp. Neurol. 286, 353 (1989). (10.1002/cne.902860306) / J. Comp. Neurol. by Barbas H. (1989)
  13. Fuster J. M., Bauer R. H., Jervey J. P., Exp. Neurol. 77, 679 (1982); (10.1016/0014-4886(82)90238-2) / Exp. Neurol. by Fuster J. M. (1982)
  14. 10.1126/science.8316836
  15. ; S. M. Courtney L. G. Ungerleider J. V. Haxby Cerebral Cortex 6 39 (1996). (10.1093/cercor/6.1.39)
  16. Eye position was monitored by means of the scleral search coil method [
  17. Robinson D. A., IEEE Trans. Biomed. Eng. 10, 137 (1963); / IEEE Trans. Biomed. Eng. by Robinson D. A. (1963)
  18. ]. The sample and test objects were presented for 400 ms and the delays were 1000 ms. The monkeys ( Macaca fasicularis and Macaca mulatta ) were required to maintain fixation of a central spot throughout the trial. At the end of the where delay the fixation spot was extinguished and small (0.3°) spots appeared simultaneously at each of the four locations used in this task (Fig. 1). This was the monkey’s “go” signal. It then had to make a direct saccade to the spot that was at the location where the match had appeared. The four locations were on the horizontal and vertical meridians 4° above below to the right and to the left of fixation. The four objects were color pictures of “real world” objects 2° by 2° in size. Each object was used as a sample or match on some trials and as a nonmatch on other trials and each object was used to cue each location. With this design what and where are linked within a single trial but across trials they are varied independently (that is no object is exclusively linked with a location). The same four objects and locations were used throughout the experiment.
  19. Recording sites were localized by magnetic resonance imaging.
  20. Delay activity was analyzed over the last 800 ms of a 1000-ms delay. We did not include the first portion of the delay to exclude any responses related to the offset of the preceding stimulus. On average data from about 600 correctly performed trials (about 85% of the total trials) were collected from each cell.
  21. We did not distinguish between location-tuned activity related to retaining sensory information about a location and activity related to a “motor set” of the forthcoming saccade. However previous studies have found that the activity of most of the PF delay neurons is related to retaining sensory rather than motor information [
  22. Funahashi S., Chafee M. V., Goldman-Rakic P. S., Nature 365, 753 (1993); (10.1038/365753a0) / Nature by Funahashi S. (1993)
  23. DiPellegrino G., Wise S. P., J. Neurosci. 13, 1227 (1993)]. (10.1523/JNEUROSCI.13-03-01227.1993) / J. Neurosci. by DiPellegrino G. (1993)
  24. We mapped a 12 mm by 12 mm extent of the lateral PF cortex at 1-mm intervals using a grid system (Crist Instruments Damascus MD). The recording chamber was centered on the principal sulcus. Its posterior end was 2 mm behind the bow of the arcuate sulcus. The dorsal and ventral recordings extended to about just above and below the superior and inferior arcuate sulci respectively. The dorsolateral PF cortex lies in and dorsal to the principal sulcus and the ventrolateral PF cortex lies on the inferior convexity ventral to the principal sulcus. What cells where cells and what-and-where cells were found about equally in the dorsolateral and ventrolateral PF cortex. In the dorsolateral PF cortex 3 what cells 25 where cells and 34 what-and-where cells were found. In the ventrolateral PF cortex 5 what cells 26 where cells and 30 what-and-where cells were found. The ANOVAs identified highly selective neurons; across the population selectivity for what and for where varied along a continuum from nonselective neurons to highly selective neurons.
  25. To eliminate any optimistic bias in the classification we performed the discriminant analysis with cross-validation; that is the distribution of means for each class was computed on half the data chosen randomly and these means were used to classify the objects or locations in the other half of the data.
  26. Because the activity of individual neurons is “noisy ” they rarely perform as well as the animal as a whole. By pooling the activity of multiple neurons it is possible to reduce noise and a neural classification rate equal to behavioral performance can thus be achieved [for details see
  27. Miller E. K., Li L., Desimone R., J. Neurosci. 13, 1460 (1993)]. (10.1523/JNEUROSCI.13-04-01460.1993) / J. Neurosci. by Miller E. K. (1993)
  28. Their relatively low classification rate for objects on the basis of where delay activity (28.3%) was not due to the generally poor ability of these neurons at classifying objects; their mean classification rate for objects on the basis of what delay activity was significantly greater (32.3% P < 0.001). The lower rate could have been due in principle to an effect of the nonmatching object which was not included in the classification and thus could add “noise.” However the mean classification rate of a discriminant analysis of the where delay activity that attempted to classify the nonmatching stimulus (24.8%) was not different from chance ( P = 0.680) indicating no effect of the nonmatch on the where delay activity. Thus the effect of object on the where delay activity appears to be due to the match object alone.
  29. See J. M. Fuster Memory in the Cerebral Cortex (MIT Press Cambridge MA 1995).
  30. Bullier J., Schall J. D., Morel A., Behav. Brain Res. 76, 89 (1996). (10.1016/0166-4328(95)00182-4) / Behav. Brain Res. by Bullier J. (1996)
  31. Ferrera V. P., Rudolph K. K., Maunsell J. H. R., J. Neurosci. 14, 6171 (1994); (10.1523/JNEUROSCI.14-10-06171.1994) / J. Neurosci. by Ferrera V. P. (1994)
  32. ; A. B. Sereno and J. H. R. Maunsell Invest. Opthal. Visual Sci. 36 692 (1995).
  33. A neural network model also suggests that the extent to which what and where are segregated or integrated in the prefrontal cortex depends on task demands [T. S. Braver and J. D. Cohen Proc. Cognit. Neurosci. Soc. 2 95 (1995)].
  34. Recanzone G. H., Merzenich M. M., Jenkins W. M., Kamil A. G., Dinse H. R., J. Neurophysiol. 67, 1057 (1992). (10.1152/jn.1992.67.5.1057) / J. Neurophysiol. by Recanzone G. H. (1992)
  35. Animal experiments were conducted in accordance with the MIT Committee on Animal Care. This work was supported by the Pew Charitable Trusts and the McKnight Foundation. We thank M. Histed for help in animal training and W. Asaad R. Desimone Y. Munakata J. Mazer R. O’Reilly and M. Wicherski for valuable comments. S.C.R. was supported by a fellowship from the McDonnell-Pew Foundation and the Markey Foundation.
Dates
Type When
Created 23 years, 1 month ago (July 27, 2002, 5:45 a.m.)
Deposited 1 year, 7 months ago (Jan. 13, 2024, 12:54 a.m.)
Indexed 3 days, 4 hours ago (Aug. 30, 2025, 12:29 p.m.)
Issued 28 years, 4 months ago (May 2, 1997)
Published 28 years, 4 months ago (May 2, 1997)
Published Print 28 years, 4 months ago (May 2, 1997)
Funders 0

None

@article{Rao_1997, title={Integration of What and Where in the Primate Prefrontal Cortex}, volume={276}, ISSN={1095-9203}, url={http://dx.doi.org/10.1126/science.276.5313.821}, DOI={10.1126/science.276.5313.821}, number={5313}, journal={Science}, publisher={American Association for the Advancement of Science (AAAS)}, author={Rao, S. Chenchal and Rainer, Gregor and Miller, Earl K.}, year={1997}, month=may, pages={821–824} }