Complex Sound Discrimination Abilities in Rats and the Effects of Multiple Training Manipulations
A.C. Puckett, C.T. Novitski, N.D. Engineer, A.L. McMenamy, M.S. Perry, C.A. Perez, P. Kan, Y.H. Chen, V. Jakkamsetti, C.L. Heydrick, M.P. Kilgard, The University of Texas at Dallas, Richardson, TX
SEQUENCE DISCRIMINATION SPEECH DISCRIMINATION
A) Place of Articulation
Dad Bad Gad
Task Target Distracter TARGET DISTRACTER Speech Stimulus Creation
Simple Frequency Discrimination Frequency Discrimination Threshold (CS+) (CS-) (CS+) (CS-)
The current experiments have 100
Example Time Course of Training
100 n.s. B) Voice Onset Time
•Recording: Monosyllabic words spoken by female native
100 90 A) Frequency
examined the ability of rats to perform 90
English speaker in a sound-proof chamber.
80 L L L H H H
–The words ‘dad’ and ‘tad’ were recorded from 5 other
discriminations among complex sounds, 70 Silence
70 native English speakers (3 male, 2 female) to assess
70 LLL B) Sequence element C) VOT Compression
including tone-noise sequences and HHH 60 discrimination H L N H L N •Frequency shifting: The frequency of the fundamental and
Dad 10% Dad 50% Dad Tad Tad 50% Tad 10%
60 50 all other formants were shifted into the rat’s hearing range by
speech stimuli. Understanding how 50
30 40 -
C) Triplet distractor doubling their frequency.
normal animals are able to discriminate 40
High first H L N H H H L L L N N N –Compressed versions of ‘dad’ and ‘tad’ were generated to
D) VOT Multiple Talker assess temporal generalization.
complex sounds is necessary in order to 30 0 10 20 30
Days of Training
40 50 60
20 E) Triplet distractor
H L N
Dad Male 1 Dad Female 2 Dad Female 1 Tad Female 1 Tad Female 2 Tad Male 1 •Noise reduction: Background noise was subtracted from
10 N N N L L L H H H
fully understand the auditory cortex each signal.
•Filtering: Each signal was filtered to correct for the
and how injury, learning or plasticity
F) Reverse Order
10 frequency-response curve of the booth speaker.
H L N N L H
can change perception and cortical Dad Dead Dud Deed Dood •Intensity adjustment: The RMS-values of the signals were
LLL (5 kHz) HHH (12 kHz) CS- Frequency (octaves above CS+) adjusted so that the loudest 100 ms of each vowel was 60 dB
F) Incremental Training
functioning. N N N N L N N N H N L H
Tone Sequence H L N
225 ms F) Glides
Behavioral data was collected from HLN is Discriminable From its Component Segments
Early Training on NNN Hinders Later Discrimination of HHH
HLN is Discriminable from Triplets of Each Segment Stop Consonants are Easily Discriminable Voice Onset Time Differences are Discriminable Good Generalization to the Majority of Compressed Variants
100 100 100 100
46 rats over 4661 total daily 90 90 HHH 90 NNN 90 90 90
n.s. LLL LLL
training sessions. 80 80
80 80 80
70 70 70
HLN 70 HLN 70 70
60 TRAINING DAYS
TRAINING DAYS 60 60 60
Methods 40 40
All animals performed Go/ No-Go discrimination tasks in 30 30
30 30 30 20
the same operant training booths. The stimuli used 20 20
20 20 20 10
as CS+ and CS- varied among different training
tasks, but the general timeline of training and 10 10 10
procedures were the same for all tasks.
HLN H L N HLN HHH LLL NNN
HLN HHH LLL NNN dad bad gad dad tad
% of original stimulus length
• Sounds are delivered from
Apparatus a speaker placed outside the
cage. Speaker is positioned
H so that sounds are usually HLN is Indistinguishable from the Reversed Sequence Enrichment Does Not Improve Reversed Sequence Discrimination Incremental Training Enhances Reversed Sequence Discrimination Generalization Across Talker and Gender Similar Vowels are Difficult to Discriminate Rats Cannot Discriminate Spectrally Similar Glides
100 100 n.s. 100 Incremental Time Course
delivered to the animal’s left n.s.
100silence 100 100 100 n.s.
ear. 90 90 90 NLN
90 90 90
• L – Rat responds to sound
80 80 80 80
stimuli by pressing on the
lever. Only presses within 3 70 70 70
40 60 70 70
CS + sounds
seconds of CS+ stimuli are
CS - sounds 30
60 60 60 50 60 60
50 50 10 40
• P – Pellet dispenser (MED 50 50 50
0 10 20 30 40 50 60 70 80 30
Associates) located outside 40 40 40 Days of Training
the sound-proof chamber
30 30 30
delivers a 45 mg food reward 10 30 30
(Bio-Serv) to the rat after 20 20 20 20 20
d M le 2
em le 3
Fe a l
• H – House light is
extinguished after false
HLN NLH HLN NLH HLN NLH dad dead dud deed dood rad lad
alarms or late responses
HLN NLH Rad Lad
Stages of Training Conclusions Conclusions
Summary: Ranked D-Prime Performance on all Sequence Tasks Summary: Ranked D-Prime Performance on all Speech Tasks
• Frequency discriminations are easy, but other sequence • Large spectral differences are easy for rats to discriminate, even if differences are not in
Shaping discriminations are difficult. the onset of the speech sound.
Goal: Animals learn to press lever to receive food • Onsets are the most salient elements of sequences. 3 3 • Onsets/consonants: /dad/ vs. /bad/ & /gad/
Time Course: ~ 6 sessions (3 days). Continued • Sequences beginning with the same element are • Middle of stimulus/vowels: /dad/ vs. /deed/ & /dood/
until 3 sessions in which animals retrieved 50 difficult to discriminate. • Subtle spectral differences are difficult for rats to discriminate (and are difficult for
• Reversed sequence discrimination was impossible, 2 2 humans as well).
despite different initial elements. • Onsets/consonants: /rad/ vs. /lad/
Detection Training •Sequences may be normally processed as a ‘unit’ • Middle of stimulus/vowels: /dad/ vs. /dead/ & /dud/
Goal: Animals learn to press to receive rewards 1 1
only after they hear a sound. Animals must learn to rather than discrete elements. • Rats can generalize among several variants of a speech stimulus.
avoid hitting during silent periods. • Discrimination strategies may be changed by training. • Rats could generalize across several compressed exemplars, indicating that VOT
Time course: ~30 sessions (3 weeks). Continued
until 10 sessions in which animals respond
• If rats learn a poor strategy early in training, they 0 0 wasn’t the only cue.
significantly more to CS+ trials than to catch trials will not learn to discriminate effectively. • Rats could generalize across several different speakers. Rats were able to generalize
- f tad
(ie, d’ >= 1.5 for 10 sessions)
• Intermediate exemplars allowed animals to adopt
ta . la
across many different fundamental frequencies, temporal patterns of enunciation, and
• Speech sounds seem to be more easily discriminated than tone-noise sequences.
Goal: Animals learn to press to receive rewards
only after they hear a CS+. Animals must avoid
hitting after CS- stimuli or during silent periods. HLN vs.
Time course: Variable (20 – 300 sessions).
Sequence Discrimination: Animals trained for Future Directions – www.utd.edu/~kilgard In Loving Memory of Matt Perry (March 1981 - September 2005)
up to 300 sessions or until behavioral criteria were
• Assessment of changes in perceptual abilities after NB-stimulation pairing Amanda Puckett In honor of the neuroscientist who delved into the mysteries and whims of life with wholehearted delight.
Speech Discrimination: Animals progressed • Measurement of changes in auditory cortex after long-term sequence training Dr. Navzer Engineer Thank you to one who touched the minds he sought to understand. As a true inquirer into philosophy,
through a variety of CS- conditions after 2 weeks of • Measurement of responses of auditory cortex after long-term speech training Crystal Novitski pharmacology and literature, Mat enriched rats as well as people.
training on each task, regardless of behavioral It is an honor to have shared a common path with you.
performance. • Assessment of cortical processing of speech sounds after environmental enrichment Vikram Jakkamsetti
• Assessment of loss of perceptual abilities after cortical injury Dr. Owen Floody Acknowledgements
• Information theory analysis of cortical responses to speech sounds Helen Chen We wish to thank all the members of the Kilgard Lab behavioral team. Research supported by NIH R21#1R15DC00662401