446:, at a precise location in the visual field (the scene)". It has been experimentally determined, for example, after mapping the angular position of some objects in the visual field, there will be a one-to-one correspondence of cells in the scene to the angular positions of those objects. Hawkins predicts that when the features of a visual scene are known in a memory, anticipatory cells should fire
36:
344:) of patterns and developing invariant representations. Higher levels of the cortical hierarchy predict the future on a longer time scale, or over a wider range of sensory input. Lower levels interpret or control limited domains of experience, or sensory or effector systems. Connections from the higher level states predispose some selected transitions in the lower-level state machines.
491:
will be active during a learned sequence. Hawkins posits that these cells will remain active for the duration of the learned sequence, even if the remainder of the cortical column is shifting state. Since we do not know the encoding of the sequence, we do not yet know the definition of
360:
is a basic element in the framework. Hawkins places particular emphasis on the role of the interconnections from peer columns, and the activation of columns as a whole. He strongly implies that a column is the cortex's physical representation of a state in a state machine.
314:
Hawkins' basic idea is that the brain is a mechanism to predict the future, specifically, hierarchical regions of the brain predict their future input sequences. Perhaps not always far in the future, but far enough to be of real use to an organism. As such, the brain is a
372:. For example, for the purposes of his framework, the nerve impulses can be taken to form a temporal sequence (but phase encoding could be a possible implementation of such a sequence; these details are immaterial for the framework).
525:
6. Hawkins' novel prediction is that certain cells are inhibited during a learned sequence. A class of cells in layers 2 and 3 should NOT fire during a learned sequence, the axons of these "exception cells" should fire
140:
458:
3. In layers 2 and 3, predictive activity (neural firing) should stop propagating at specific cells, corresponding to a specific prediction. Hawkins does not rule out anticipatory cells in layers 4 and 5.
935:
368:, but merely signals that the natural process has performed Hawkins' functional decomposition in a different, unexpected way, as Hawkins' motivation is to create intelligent
17:
364:
As an engineer, any specific failure to find a natural occurrence of some process in his framework does not signal a fault in the memory-prediction framework
475:
are in layer 2, physically adjacent to layer 1. Hawkins does not rule out the existence of layer 3 cells with dendrites in layer 1, which might perform as
686:
940:
826:
538:
7. If an unusual event occurs (the learned sequence fails), the "exception cells" should fire, propagating up the cortical hierarchy to the
770:
George, Dileep; Hawkins, Jeff (2005). "A Hierarchical
Bayesian Model of Invariant Pattern Recognition in the Visual Cortex": 1812–1817.
895:
669:
685:
Fogassi, Leonardo; Ferrari, Pier
Francesco; Gesierich, Benno; Rozzi, Stefano; Chersi, Fabian; Rizzolatti, Giacomo (April 29, 2005).
233:
119:
882:
53:
930:
100:
350:
is part of the framework, in which the event of learning physically alters neurons and connections, as learning takes place.
57:
72:
619:
255:
550:
8. Hawkins predicts a cascade of predictions, when recognition occurs, propagating down the cortical column (with each
79:
808:- Research papers and programs presenting experimental results with Bayesian models of the Memory-Prediction Framework
625:
309:
293:
319:
805:
86:
910:
388:
as a prototype for some example predictions, such as
Predictions 2, 8, 10, and 11. Other predictions cite the
133:
On
Intelligence: How a New Understanding of the Brain will Lead to the Creation of Truly Intelligent Machines
281:
On
Intelligence: How a New Understanding of the Brain will Lead to the Creation of Truly Intelligent Machines
18:
On
Intelligence: How a New Understanding of the Brain will Lead to the Creation of Truly Intelligent Machines
68:
46:
530:. This prevents flooding the brain with the usual sensations, leaving only exceptions for post-processing.
771:
334:
316:
785:
733:
595:
701:
776:
850:
725:
353:
471:. Hawkins calls the cells which fire in this sequence "name cells". Hawkins suggests that these
717:
665:
661:
654:
622:, a technology by Hawkins's startup Numenta Inc. to replicate the properties of the neocortex.
582:. Hawkins posits a temporal window (presuming time-encoded firing) which is necessary for his
341:
240:
228:
93:
903:
867:
842:
709:
347:
289:
266:
157:
798:
509:
439:
406:
389:
357:
705:
599:
567:
463:"Name cells" at layers 2 and 3 should preferentially connect to layer 6 cells of cortex
886:
924:
443:
422:
385:
330:
914:
729:
610:
11. Hawkins predicts that "name cells" will be found in all regions of the cortex.
285:
153:
562:
Pyramidal cells should detect coincidences of synaptic activity on thin dendrites
539:
504:(i.e., the name cells simultaneously "light up") across an array of name cells.
501:
185:
35:
442:, Hawkins predicts, for example, "we should find anticipatory cells in or near
337:
state machine, the machine responds to future events predicted from past data.
414:
381:
175:
811:
713:
555:
213:
721:
454:
Prediction should stop propagating in the cortical column at layers 2 and 3
500:; Hawkins suggests that the ON pattern may be as simple as a simultaneous
571:
323:
247:
871:
846:
590:
Learned representations move down the cortical hierarchy, with training
579:
551:
369:
340:
The hierarchy is capable of memorizing frequently observed sequences (
583:
575:
687:"Parietal lobe: from action organization to intention understanding"
297:
814:- An open source project for modeling Memory-Prediction Framework
838:
762:
139:
858:
Colwell, B. (2005). "Machine
Intelligence Meets Neuroscience".
333:
actually controls the behavior of the organism. Since it is a
396:
An
Appendix of 11 Testable Predictions, beginning on page 237:
29:
521:"Exception cells" should remain OFF during a learned sequence
467:
4. Learned sequences of firings comprise a representation of
241:
598:(IT) level has learned a sequence, that eventually cells in
376:
Predictions of the theory of the memory-prediction framework
401:
Enhanced neural activity in anticipation of a sensory event
267:
570:
should be capable of detecting coincident events on thin
534:"Exception cells" should propagate unanticipated events
483:"Name cells" should remain ON during a learned sequence
265:
253:
239:
227:
219:
209:
201:
191:
181:
171:
163:
149:
60:. Unsourced material may be challenged and removed.
653:
413:", cells that fire in anticipation of a sensory
936:Non-fiction books about Artificial intelligence
546:"Aha! cells" should trigger predictive activity
594:10. Hawkins posits, for example, that if the
8:
132:
606:"Name cells" exist in all regions of cortex
450:the actual objects are seen in the scene.
409:, Hawkins (2004) predicts "we should find
322:with special properties that enable it to
138:
131:
827:"Machine Intelligence Meets Neuroscience"
775:
647:
645:
643:
641:
120:Learn how and when to remove this message
300:and describes some of its consequences.
896:"On Intelligence, People and Computers"
637:
911:On Biological and Digital Intelligence
794:
783:
660:(1st ed.). Times Books. pp.
528:only if a local prediction is failing
7:
558:over a learned scene, for example).
516:which perform this type of function.
58:adding citations to reliable sources
806:Saulius Garalevicius' research page
894:Kling, Arnold (22 November 2004).
542:, the repository of new memories.
25:
881:Dill, Franz (October 30, 2004).
34:
883:"Jeff Hawkins: On Intelligence"
392:( Predictions 1, 3, 4, and 7).
45:needs additional citations for
941:Books about human intelligence
602:will also learn the sequence.
469:temporally constant invariants
1:
825:Colwell, Bob (January 2005).
489:temporally constant invariant
434:Spatially specific prediction
292:. The book explains Hawkins'
620:Hierarchical temporal memory
626:Memory-prediction framework
425:have been observed to fire
310:Memory-prediction framework
294:memory-prediction framework
957:
320:hierarchical state machine
307:
137:
510:Neural ensemble#Encoding
714:10.1126/science.1106138
931:2004 non-fiction books
793:Cite journal requires
652:Hawkins, Jeff (2004).
438:2. In primary sensory
429:an anticipated event.
900:Tech Central Station
487:5. By definition, a
356:'s formulation of a
54:improve this article
27:Book by Jeff Hawkins
706:2005Sci...308..662F
514:grandmother neurons
405:1. In all areas of
134:
872:10.1109/MC.2005.24
847:10.1109/MC.2005.24
586:to remain viable.
578:with thousands of
411:anticipatory cells
354:Vernon Mountcastle
284:is a 2004 book by
812:Project Neocortex
700:(5722): 662–667.
421:Note: As of 2005
342:Cognitive modules
277:
276:
202:Publication place
130:
129:
122:
104:
69:"On Intelligence"
16:(Redirected from
948:
907:
902:. Archived from
890:
885:. Archived from
875:
854:
849:. Archived from
802:
796:
791:
789:
781:
779:
766:
765:
763:Official website
748:
747:
745:
744:
738:
732:. Archived from
691:
682:
676:
675:
659:
649:
348:Hebbian learning
290:Sandra Blakeslee
273:QP376 .H294 2004
269:
243:
193:Publication date
158:Sandra Blakeslee
142:
135:
125:
118:
114:
111:
105:
103:
62:
38:
30:
21:
956:
955:
951:
950:
949:
947:
946:
945:
921:
920:
893:
880:
857:
824:
821:
792:
782:
777:10.1.1.132.6744
769:
761:
760:
757:
752:
751:
742:
740:
736:
689:
684:
683:
679:
672:
656:On Intelligence
651:
650:
639:
634:
616:
608:
592:
568:Pyramidal cells
564:
548:
536:
523:
485:
465:
456:
436:
403:
390:auditory system
378:
358:cortical column
312:
306:
258:
210:Media type
194:
145:
126:
115:
109:
106:
63:
61:
51:
39:
28:
23:
22:
15:
12:
11:
5:
954:
952:
944:
943:
938:
933:
923:
922:
919:
918:
908:
906:on 2012-03-05.
891:
889:on 2012-02-05.
878:
877:
876:
853:on 2005-02-04.
820:
817:
816:
815:
809:
803:
795:|journal=
767:
756:
755:External links
753:
750:
749:
677:
671:978-0805074567
670:
636:
635:
633:
630:
629:
628:
623:
615:
612:
607:
604:
596:inferotemporal
591:
588:
563:
560:
547:
544:
535:
532:
522:
519:
518:
517:
484:
481:
464:
461:
455:
452:
435:
432:
431:
430:
423:mirror neurons
402:
399:
398:
397:
377:
374:
308:Main article:
305:
302:
296:theory of the
275:
274:
271:
263:
262:
259:
254:
251:
250:
245:
237:
236:
231:
225:
224:
221:
217:
216:
211:
207:
206:
203:
199:
198:
195:
192:
189:
188:
183:
179:
178:
173:
169:
168:
165:
161:
160:
151:
147:
146:
143:
128:
127:
42:
40:
33:
26:
24:
14:
13:
10:
9:
6:
4:
3:
2:
953:
942:
939:
937:
934:
932:
929:
928:
926:
916:
912:
909:
905:
901:
897:
892:
888:
884:
879:
873:
869:
865:
861:
856:
855:
852:
848:
844:
840:
836:
832:
828:
823:
822:
818:
813:
810:
807:
804:
800:
787:
778:
773:
768:
764:
759:
758:
754:
739:on 2017-08-09
735:
731:
727:
723:
719:
715:
711:
707:
703:
699:
695:
688:
681:
678:
673:
667:
663:
658:
657:
648:
646:
644:
642:
638:
631:
627:
624:
621:
618:
617:
613:
611:
605:
603:
601:
597:
589:
587:
585:
581:
577:
574:, even for a
573:
569:
561:
559:
557:
553:
545:
543:
541:
533:
531:
529:
520:
515:
511:
507:
506:
505:
503:
499:
495:
490:
482:
480:
478:
474:
470:
462:
460:
453:
451:
449:
445:
441:
433:
428:
424:
420:
419:
418:
416:
412:
408:
400:
395:
394:
393:
391:
387:
386:visual system
383:
375:
373:
371:
367:
362:
359:
355:
351:
349:
345:
343:
338:
336:
332:
331:state machine
327:
325:
321:
318:
311:
303:
301:
299:
295:
291:
287:
283:
282:
272:
270:
268:LC Class
264:
260:
257:
256:Dewey Decimal
252:
249:
246:
244:
238:
235:
234:0-8050-7456-2
232:
230:
226:
222:
218:
215:
212:
208:
205:United States
204:
200:
196:
190:
187:
184:
180:
177:
174:
170:
166:
162:
159:
155:
152:
148:
141:
136:
124:
121:
113:
102:
99:
95:
92:
88:
85:
81:
78:
74:
71: –
70:
66:
65:Find sources:
59:
55:
49:
48:
43:This article
41:
37:
32:
31:
19:
917:(7 Oct 2004)
915:Ben Goertzel
913:A review by
904:the original
899:
887:the original
863:
859:
851:the original
834:
830:
786:cite journal
741:. Retrieved
734:the original
697:
693:
680:
655:
609:
593:
565:
549:
537:
527:
524:
513:
497:
493:
488:
486:
476:
472:
468:
466:
457:
447:
437:
426:
410:
404:
379:
365:
363:
352:
346:
339:
335:feed forward
328:
317:feed forward
313:
286:Jeff Hawkins
280:
279:
278:
154:Jeff Hawkins
116:
107:
97:
90:
83:
76:
64:
52:Please help
47:verification
44:
540:hippocampus
382:predictions
186:Times Books
144:Front cover
925:Categories
743:2006-11-18
632:References
477:name cells
473:name cells
304:The theory
261:612.8/2 22
176:Psychology
110:April 2013
80:newspapers
866:: 12–15.
841:: 12–15.
772:CiteSeerX
572:dendrites
214:Paperback
182:Publisher
860:Computer
831:Computer
722:15860620
614:See also
580:synapses
384:use the
370:machines
248:55510125
164:Language
819:Reviews
730:5720234
702:Bibcode
694:Science
554:of the
552:saccade
172:Subject
167:English
94:scholar
774:
728:
720:
668:
584:theory
576:neuron
498:active
448:before
440:cortex
427:before
407:cortex
366:per se
156:&
150:Author
96:
89:
82:
75:
67:
837:(1).
737:(PDF)
726:S2CID
690:(PDF)
415:event
324:learn
298:brain
220:Pages
101:JSTOR
87:books
839:IEEE
799:help
718:PMID
666:ISBN
512:for
508:See
380:His
329:The
288:and
242:OCLC
229:ISBN
197:2004
73:news
868:doi
843:doi
710:doi
698:308
662:272
566:9.
556:eye
502:AND
496:or
223:272
56:by
927::
898:.
864:38
862:.
835:38
833:.
829:.
790::
788:}}
784:{{
724:.
716:.
708:.
696:.
692:.
664:.
640:^
600:V4
494:ON
479:.
444:V1
417:.
326:.
874:.
870::
845::
801:)
797:(
780:.
746:.
712::
704::
674:.
123:)
117:(
112:)
108:(
98:·
91:·
84:·
77:·
50:.
20:)
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.
