[AUDITORY] Maximum temporal window for pattern recognition and the perception of redundancy (Steven Greenberg )

Subject: [AUDITORY] Maximum temporal window for pattern recognition and the perception of redundancy From: Steven Greenberg <steveng@xxxxxxxx> Date: Sun, 5 Feb 2017 12:52:24 -0800 List-Archive:<http://lists.mcgill.ca/scripts/wa.exe?LIST=AUDITORY> This is a multi-part message in MIME format. --------------010609040805030408020908 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by edgeum2.it.mcgill.ca id v15Kqxg6020433 Dear Dario, The term =E2=80=9Cwindow=E2=80=9D may be inappropriate in this context, a= s it implies a=20 =E2=80=9Cbrick wall=E2=80=9D type of temporal filter. A more appropriate = metaphor may be=20 =E2=80=9Ctime constant=E2=80=9D with an exponential decay function. The p= recise function=20 depends on the nature of the sensory input and the purpose to which it=E2= =80=99s=20 put. Erickson=E2=80=99s visual integration study of ca. 1965 suggests that eid= etic=20 imagery (a form of visual memory) may persist up to ca. 1.5 seconds. He=20 used quasi-random dot patterns, which when presented concurrently to=20 both visual fields, formed tri-gram letter sequences. By varying the=20 time delay between the two dot patterns, the decay function (for this=20 task) could be deduced. Experiments on tonal fission and fusion (by van Noorden, Bregman and=20 others) can be interpreted similarly (translating frequency distance=20 into appropriate spatial coordinates). In speech, the average duration of a =E2=80=9Cturn=E2=80=9D in spontaneou= s dialogue is=20 ca. 2 seconds (it varies somewhat, depending on speaking style, topic=20 and communication medium). Integration of linguistic elements of=20 variable scale (phonetic feature, segment, syllable, word, phrase) into=20 =E2=80=9Cmeaning=E2=80=9D is likely to conform to comparable limits as th= ose cited above. Cowan and others have estimated the span of short-term memory to be ca.=20 4 seconds, an interval close to the typical upper limit of a talker=E2=80= =99s=20 =E2=80=9Cturn=E2=80=9D in spontaneous (true) dialogues. The sensory systems are closely intertwined with the memory systems,=20 which in turn are intimately associated with communication, emotion and=20 action. This is one reason why estimates of time constants/windows vary=20 somewhat across modality, task and subject. Best wishes, and good luck with your research. Steven Greenberg --------------010609040805030408020908 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by edgeum2.it.mcgill.ca id v15Kqxg6020433 <html><head> <meta content=3D"text/html; charset=3DUTF-8" http-equiv=3D"Content-Type"> </head><body style=3D"font-family: Helvetica,Arial,sans-serif; font-size:= =20 12pt;" text=3D"#000000" bgcolor=3D"#FFFFFF"> <div style=3D"font-size: 12pt;font-family: Helvetica,Arial,sans-serif;"><= p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt;line-height:115%"><spa= n=20 style=3D"font-family:Helvetica">Dear Dario,<o:p></o:p> <s= pan style=3D"font-family:Helvetica">The term =E2=80=9Cwindow=E2=80=9D may be= inappropriate in this context, as it implies a =E2=80=9Cbrick wall=E2=80=9D type of temporal fi= lter. A more appropriate metaphor may be =E2=80=9Ctime constant=E2=80=9D with an expon= ential decay=20 function. The precise function depends on the nature of the sensory input and the=20 purpose to which it=E2=80=99s put.<o:p></o:p> <s= pan style=3D"font-family:Helvetica">Erickson=E2=80=99s visual integration st= udy of=20 ca. 1965 suggests that eidetic imagery (a form of visual memory) may persist up to ca. <spa= n __postbox-detected-content=3D"__postbox-detected-date"=20 class=3D"__postbox-detected-content __postbox-detected-date"=20 style=3D"display: inline; font-size: inherit; padding: 0pt;">1.5 seconds. He used quasi-random dot patterns, which when presented=20 concurrently to both visual fields, formed tri-gram letter sequences. By varying the=20 time delay between the two dot patterns, the decay function (for this task) could=20 be deduced. <o:p></o:p> <s= pan style=3D"font-family:Helvetica">Experiments on tonal fission and fusion=20 (by van Noorden, Bregman and others) can be interpreted similarly (translating frequency distance into appropriate spatial coordinates).<o:p></o:p></spa= n> <s= pan style=3D"font-family:Helvetica">In speech, the average duration of a=20 =E2=80=9Cturn=E2=80=9D in spontaneous dialogue is ca. 2 seconds (it varies somewhat, depending on=20 speaking style, topic and communication medium). Integration of linguistic=20 elements of variable scale (phonetic feature, segment, syllable, word, phrase) into =E2=80=9Cmeaning=E2=80=9D is likely to conform to comparable limits as th= ose cited=20 above.<o:p></o:p> <s= pan style=3D"font-family:Helvetica">Cowan and others have estimated the span= =20 of short-term memory to be ca. 4 seconds, an interval close to the typical=20 upper limit of a talker=E2=80=99s =E2=80=9Cturn=E2=80=9D in spontaneous (true) dialog= ues.<o:p></o:p> = The sensory systems are closely=20 intertwined with the memory systems, which in turn are intimately associated with=20 communication, emotion and action. This is one reason why estimates of time=20 constants/windows vary somewhat across modality, task and subject. <p=20 class=3D"MsoNormal" style=3D"margin-bottom: 12pt; line-height: 115%;"><sp= an=20 style=3D"font-family: Helvetica;">Best wishes, and good luck with your=20 research. Steven Greenberg</div> </body> </html> --------------010609040805030408020908--

This message came from the mail archive
../postings/2017/
maintained by:

DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University